0 Голоса «за»0 Голоса «против»

Просмотров: 2551 стр.Feb 22, 2018

© © All Rights Reserved

PDF, TXT или читайте онлайн в Scribd

© All Rights Reserved

Просмотров: 2

© All Rights Reserved

- Army Aviation Digest - May 1965
- Gyro Plane
- Differential Transform Method for Solving Engineering and Mathematical Problems Version 5.pdf
- FM 1-514 Principles of Helicopter Flight
- Development Of Helicopters
- Paper 2
- Robinson R44 IPC
- Rigid-Flexible Contact Analysis of an Inflated Mem
- Aerospace syllabus
- 2015 Webinar Implicit
- Sahu
- Rotorcraft Simulation Modelling
- ACC08_Promtun
- Blow-Up of Finite Difference Solutions to Parabolic Equations With Semilinear Dynamical Boundary Conditions - Miglena Koleva and Lubin Vulkov
- MAE4242_Ch05_Directional Static Stability.pdf
- Existence and Stability
- Mae 331 Lecture 5
- 2012o%27neillmscr
- Advanced Non-linear Control Algorithms Applied to Design Highly
- CHEE319 Quiz1 2015 Solu

Вы находитесь на странице: 1из 551

SIMULATION WITH A TIME-ACCURATE

FREE-VORTEX WAKE MODEL

Department of Aerospace Engineering

model has been used to analyze different flight conditions, including both steady

includes flexible blades, with coupled flap-lag-torsion dynamics and swept tips; the

rigid body dynamics are modeled with the non-linear Euler equations. The free

wake models the rotor flow field by tracking the vortices released at the blade tips.

imated using finite differences, and solved using a time-accurate numerical scheme.

The flight dynamics model can be solved as a system of non-linear algebraic trim

to pilot-applied controls. This study also implements new approaches to reduce the

prohibitive computational costs associated with such complex models without losing

accuracy.

The mathematical model was validated for trim conditions in level flight, turns,

climbs and descents. The results obtained correlate well with flight test data, both

in level flight as well as turning and climbing and descending flight. The swept tip

model was also found to improve the trim predictions, particularly at high speed.

The behavior of the rigid body and the rotor blade dynamics were also studied and

related to the aerodynamic load distributions obtained with the free wake induced

velocities.

The model was also validated in a lateral maneuver from hover. The results

show improvements in the on-axis prediction, and indicate a possible relation be-

tween the off-axis prediction and the lack of rotor-body interaction aerodynamics.

The swept blade model improves both the on-axis and off-axis response. An ax-

ial descent though the vortex ring state was simulated. As the“vortex ring” goes

through the rotor, the unsteady loads produce large attitude changes, unsteady flap-

ping, fluctuating thrust and an increase in power required. A roll reversal maneuver

specifically the effect of the aerodynamic loading on the rotor orientation and the

off-axis response.

Helicopter Flight Dynamics Simulation

With a Time-Accurate Free-Vortex

Wake Model

by

Maria Ribera

University of Maryland, College Park in partial fulfillment

of the requirements for the degree of

Doctor of Philosophy

2007

Advisory Commmittee:

Professor James Baeder

Professor Inderjit Chopra

Professor J. Gordon Leishman

Professor Tobias von Petersdorff

c Copyright by

Maria Ribera

2007

To my parents, for giving me wings,

and to my husband, for flying with me.

y a mi marido, por volar a mi lado.

ii

ACKNOWLEDGMENTS

but he also opened my mind to the possibility of rotorcraft, a field both challenging

and exciting, and now I cannot understand why I never thought of it before.

James Baeder, Dr. Inderjit Chopra and Dr. von Petersdorff for the guidance and

support. Particularly, to Dr. J. Gordon Leishman, for sharing with me with the

and Carlos Malpica, I owe them much for their help and useful discussions. They

offered me a warm welcome and their friendship. Our many years sharing the cubicle

and the code have made Carlos and I much more than coworkers, he and his wife

I have formed many other friendships in the Rotorcraft Center, the list is too

long, but I particularly want to remember Anne, for her laughs, Tracy, for being a

girlfriend among all the boys, and Felipe, for the fun conversations and gossip. They

have all made my time here much more pleasant and easy to endure, and I thank

them for that. Most specially, I am forever indebted to Shreyas Ananthan. He has

taught me much about aerodynamics and wake models, but also about perseverance

iii

and confidence. I would not be were I am today without his help, and this work

The friends I have made at the University of Maryland have opened my eyes

to a world of diversity, without borders, and have, during these seven years, become

Sandra Pereira my sisters, I love them and thank them for sticking by me through

the thick and thin, in the good and the bad moments. We have had so much fun

together, and they truly helped me lead a balanced life. Their families have become

the Pereira family, Gustavo, Isabel, Frederico and Olga, who have adopted me and

cared for me all these years, and Linda Curtis, for her extreme kindness in giving me

a home and a listening ear in the most difficult times of this PhD, the writing year.

My good friend Wendy Abi Assali and her daughter Sabrina have also opened her

heart and her home to me towards the end of this degree. I am thankful for her trust,

her friendship and her support. She has understood like noone else the frustrations

and hardships of a long PhD, and I wish her the best of luck to complete hers soon

as well. And last, but not least, I want to thank Igor Alonso, one of my oldest

and dearest friends. We shared our college years and embarked in the graduate

adventure together. He has always been there, and I know he will always be.

To my parents and my brother, this dissertation has felt longer and more

painful than even to myself. They have endured my absence with patience and

understanding, with constant reminders of their love and support, and have never

iv

I have made them proud, and somehow that makes all the low points of this PhD

worthwhile. I love them and I will never have words to thank them enough, although

And I leave my husband Luis for the end, but he is indeed the beginning. The

end of this PhD has been marked by our wedding and first months of married life. I

the hardest tests and our love has come out stronger. And I cannot wait to discover

v

TABLE OF CONTENTS

Nomenclature xxx

1 Introduction 1

1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

2 Mathematical Model 33

2.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

vi

2.2.2 Main rotor coordinate systems . . . . . . . . . . . . . . . . . . 38

2.4.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48

3.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88

vii

3.2 Assumptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89

3.3.3 Coupling of the flight dynamic and the free wake coordinate

systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93

3.7 The free wake model with swept tip blades . . . . . . . . . . . . . . . 102

4.3.1 Assembly of mass and stiffness matrices with swept tip model 118

viii

4.5.3 The trim equations . . . . . . . . . . . . . . . . . . . . . . . . 126

4.5.5 Coupling of the free wake and rotor-fuselage models in trim . 130

4.5.6 Extraction of the state vector from the trim solution . . . . . 137

4.6.3 Response after trimming with a relaxation free wake model . . 146

ix

6.2.5 Effect of the swept tip . . . . . . . . . . . . . . . . . . . . . . 188

x

7.3.1 Rigid body response . . . . . . . . . . . . . . . . . . . . . . . 403

and a time-marching free wake model for the integrated response 484

Bibliography 490

xi

LIST OF TABLES

5.1 Main parameters of the UH-60A Black Hawk helicopter configuration. 160

xii

LIST OF FIGURES

2.2 Euler angles and rotations from the inertial to fuselage coordinate

2.7 Definition of the sectional aerodynamic angles, the yaw angle γI and

3.2 Definition of equivalent tip flapping angles for the free wake model. . 105

4.2 Blade degrees of freedom for the complete blade using four finite

elements. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148

4.3 Blade degrees of freedom for the complete blade using four finite

elements for the straight portion of the blade and one for the swept tip.149

xiii

4.4 Schematic of the assembly of the straight and swept portions of the

4.6 Scheme of the trim procedure with a time-marching free wake model

number M . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162

number M . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162

5.4 First, second and third natural mode shapes for the baseline model

5.5 Fourth, fifth and sixth natural mode shapes for the baseline model

5.6 First six natural frequencies with varying tip sweep angle. . . . . . . 166

5.7 First, second and third natural mode shapes for the model with swept

tip. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167

5.8 Fourth, fifth and sixth natural mode shapes for the model with swept

tip. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168

xiv

6.1 Main rotor power required, QM R , as a function of speed, for a straight

configuration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242

6.5 Main rotor lateral stick position as a function of speed, for a straight

configuration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244

6.8 Side view of the wake geometry at V = 1, 40 and 100 kts. . . . . . . . 246

6.9 Rear view of the wake geometry at V = 1, 40 and 100 kts. . . . . . . 247

6.10 Top view of the wake geometry at V = 1, 40 and 100 kts. . . . . . . . 248

6.12 Distribution of aerodynamic angle of attack over the rotor disk. . . . 250

6.15 Distribution of local flap moment, rCL M 2 , over the rotor disk. . . . . 253

xv

6.18 Elemental induced torque distribution, rCL M 2 sin φ, over the rotor

disk. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 256

6.19 Elemental profile torque distribution, rCD M 2 cos φ, over the rotor disk.257

6.20 Distribution of pitching moment coefficient, CM , over the rotor disk. . 258

6.25 Magnitude and phase of the first three lag harmonics at V = 1 kt, 40

6.29 Main rotor collective as a function of speed, with a rigid blade model

6.30 Comparison of the induced inflow distribution with the rigid and

xvi

6.31 Equivalent flap angle as a function of azimuth azimuth at V = 1 kt,

blade section for a longitudinal cross section of the rotor, with the

of the rotor, with the rigid and flexible blade models; V = 1 kt. . . . 271

marching free wake model with a straight and swept tip blade models. 272

6.35 Main rotor collective as a function of speed with a straight and swept

6.36 Elemental induced torque, rCL M 2 sin φ, with a straight and swept

6.37 Elemental profile torque, rCD M 2 cos φ, with a straight and swept tip

6.38 Induced velocities, λ, with a straight and swept tip blade models, at

6.39 Bound circulation, Γ, with a straight and swept tip blade models, at

6.40 Relation between the roll angle, φ, and the turn rate, ψ̇. . . . . . . . 278

xvii

6.43 Lateral stick position, δlat , and longitudinal stick position, δlon , as a

6.44 Helicopter pitch angle, θ, pitch rate, q and roll rate, p, as a function

6.45 Pedal position δped , and helicopter yaw rate, r, as a function of roll

angle φ. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283

6.46 Side view of the free wake tip vortex for turns at V = 60 kts. . . . . . 284

6.47 Rear view of the free wake tip vortex for turns at V = 60 kts. . . . . 285

6.48 Top view of the free wake tip vortex for turns at V = 60 kts. . . . . . 286

6.50 Angle of attack distribution for different turn rates at V = 60 kts. . . 288

6.51 Lift coefficient distribution for different turn rates at V = 60 kts. . . . 289

6.52 Elemental lift distribution for different turn rates at V = 60 kts. . . . 290

6.53 Local flap moment distribution for different turn rates at V = 60 kts. 291

6.54 Drag coefficient distribution for different turn rates at V = 60 kts. . . 292

6.55 Elemental drag distribution for different turn rates at V = 60 kts. . . 293

6.56 Induced torque distribution for different turn rates at V = 60 kts. . . 294

6.57 Profile torque distribution for different turn rates at V = 60 kts. . . . 295

6.58 Moment coefficient distribution for different turn rates at V = 60 kts. 296

6.59 Elemental moment distribution for different turn rates at V = 60 kts. 297

xviii

6.61 Magnitude and phase of the first three flap harmonics at various turn

6.63 Magnitude and phase of the first three lag harmonics at various turn

6.64 Blade tip elastic torsion versus azimuth angle at various turn rates;

V = 60 kts. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 302

6.65 Magnitude and phase of the first three torsion harmonics at various

6.66 Main rotor power required, QM R , and collective stick position, δcol ,

kts. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 304

6.67 Longitudinal stick position, δlon , and lateral stick position, δlat , as a

6.68 Helicopter pitch attitude, θ, and pedal setting, δped ,as a function of

6.69 Side view of the wake geometry for different climb angles γ; V = 60

kts. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307

6.70 Rear view of the wake geometry for different climb angles γ; V = 60

kts. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 308

6.71 Top view of the wake geometry for different climb angles γ; V = 60

kts. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 309

xix

6.72 Side view of the wake geometry for different descent angles γ; V = 60

kts. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 310

6.73 Rear view of the wake geometry for different descent angles γ; V = 60

kts. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 311

6.74 Top view of the wake geometry for different descent angles γ; V = 60

kts. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 312

6.77 Angle of attack distribution for different climb angles γ; V = 60 kts. . 315

6.78 Angle of attack distribution for different descent angles γ; V = 60 kts. 316

6.79 Lift coefficient distribution for different climb angles γ; V = 60 kts. . 317

6.80 Lift coefficient distribution for different descent angles γ; V = 60 kts. 318

6.81 Elemental lift distribution for different climb angles γ; V = 60 kts. . . 319

6.82 Elemental lift distribution for different descent angles γ; V = 60 kts. . 320

6.83 Local flap moment for different climb angles γ; V = 60 kts. . . . . . . 321

6.84 Local flap moment for different descent angles γ; V = 60 kts. . . . . . 322

6.85 Drag coefficient distribution for different climb angles γ; V = 60 kts. . 323

6.86 Drag coefficient distribution for different descent angles γ; V = 60 kts. 324

6.87 Elemental drag distribution for different climb angles γ; V = 60 kts. . 325

6.88 Elemental drag distribution for different descent angles γ; V = 60 kts. 326

6.89 Elemental induced torque for different climb angles γ; V = 60 kts. . . 327

6.90 Elemental induced torque for different descent angles γ; V = 60 kts. . 328

6.91 Elemental profile torque for different climb angles γ; V = 60 kts. . . . 329

xx

6.92 Elemental profile torque for different descent angles γ; V = 60 kts. . . 330

kts. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 331

kts. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 333

kts. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 334

kts. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 335

6.98 Blade tip equivalent flapping angle versus azimuth angle in climbs

6.99 Magnitude (top) and phase (bottom) of the first three flap harmonics

6.100Blade tip equivalent lag angle versus azimuth angle in climbs and

6.101Magnitude (top) and phase (bottom) of the first three lag harmonics

6.102Blade tip elastic torsion versus azimuth angle in climbs and descents;

V = 60 kts. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 340

xxi

6.104Vortex ring state boundary from the experiments of Drees and Hendal.342

6.105Main rotor power required and collective for different descent angles

γ; V = 20 kts. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 343

6.106Fuselage angle of attack and pitch angle, and main rotor longitudinal

6.107Fuselage roll angle, main rotor lateral cyclic and pedal setting for

6.109Side and top views of the wake geometry for γ = −30◦ (top), γ =

V = 20 kts. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 350

xxii

6.115Blade elastic torsion angle versus azimuth displacement for γ = −30◦ ,

6.117Lateral stick position, δlat , helicopter roll angle, φ and roll rate p, for

γ = −40◦ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 356

6.120Side view of the free wake tip vortex geometry for turn rates from

6.121Side view of the wake geometry for turn rates from ψ̇ = ±30 deg/sec

6.122Rear view of the wake geometry for turn rates from ψ̇ = 0 deg/sec to

6.123Rear view of the wake geometry for turn rates from ψ̇ = ±30 deg/sec

6.124Top view of the wake geometry for turn rates from ψ̇ = 0 deg/sec to

6.125Top view of the wake geometry for turn rates from ψ̇ = ±30 deg/sec

xxiii

6.126Inflow distribution for turn rates at ψ̇ = 0 deg/sec, ψ̇ = ±20 deg/sec

7.1 Time history of the control input, roll rate p, and roll angle φ, for a

7.2 Time history of the control input, pitch rate q, and pitch angle θ for

7.3 Time history of the control input, yaw rate r and yaw angle ψ for a

7.4 Time history of the control input and flap coefficients, β0 , β1c and β1s

xxiv

7.5 Time history of the control input and lag coefficients, ζ0 , ζ1c and ζ1s

7.6 Side (left) and rear (right) views of the wake geometry at different

7.7 Inflow λ (left) and inflow difference (since the beginning of the ma-

kt. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 422

7.11 Lift contribution to roll (left) and pitch (right) for a lateral maneuver;

V = 1 kt. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 426

7.12 Time history of the control input, roll rate p and roll angle φ for a

lateral maneuver with the swept tip modeled in the blade; V = 1 kt. . 427

7.13 Time history of the control input, pitch rate q and pitch angle θ for

a lateral maneuver with the swept tip modeled in the blade; V = 1 kt. 428

7.14 Time history of the control input, roll rate p and roll angle φ for a

xxv

7.15 Time history of the control input, pitch rate q and pitch angle θ for

7.16 Time history of the control input, yaw rate r and yaw angle ψ for a

7.17 Time history of the control input, pitch angle θ and pitch rate q for

7.18 Time history of the control input, roll angle φ and roll rate p for an

7.19 Time history of the control input, thrust coefficient CT and main

7.20 Time history of the control input and forward and vertical velocities,

7.21 Time history of the control input and the lateral velocity v yaw angle

7.22 Time history of the flap coefficients, β0 , β1c and β1s for an axial

7.23 Time history of the control input and lag coefficients, ζ0 , ζ1c and ζ1s

7.24 Time history of the control input and torsion coefficients, φ0 , φ1c and

7.25 Side view of the free wake tip vortex geometry at different times for

xxvi

7.26 Rear view of the free wake tip vortex geometry at different times for

xxvii

7.37 Lift contribution to roll moment rCL M 2 sin ψ at different times for

7.38 Lift contribution to pitch moment rCL M 2 cos ψ at different times for

7.39 Time history of the control input, roll rate p and roll angle φ for a

7.40 Time history of the control input, pitch rate q and pitch angle θ for

7.41 Time history of the control input and forward, lateral and vertical

80 kts. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 456

7.42 Time history of the control input, main rotor power required QM R

7.43 Time history of the control input and flap coefficients, β0 , β1c and β1s

7.44 Time history of the control input and lag coefficients, ζ0 , ζ1c and ζ1s

7.45 Time history of the control input and torsion coefficients, φ0 , φ1c and

7.46 Rear view of the wake geometry at different times for a roll reversal

7.47 Side view of the wake geometry at different times for a roll reversal

xxviii

7.48 Distribution of induced velocities over the rotor at different times for

7.49 Distribution of angle of attack over the rotor at different times for a

7.52 Distribution of lift contribution to roll moment rCL M 2 sin ψ, over the

the rotor at different times for a roll reversal maneuver; V = 80 kts. . 468

model to the flight dynamics simulation after obtaining the trim so-

B.2 Schematic of the free wake convergence phase between trimming with

the relaxation free wake and integrating the transient response with

C.1 Schematic of the trim procedure with the Bagai-Leishman free wake. 489

xxix

NOMENCLATURE

a Lift-curve slope

a Acceleration vector

A State matrix

b Semi-chord length

B Control matrix

êx , êy , êz Unit vectors of undeformed preconed blade reference frame

G Shear modulus

xxx

iG , jG , kG Unit vectors of global wake reference frame

i Unit vector

Ixx , Iyy , Izz Aircraft mass moments of inertia about body axes

j Unit vector

k Unit vector

M Mach number

n Load factor

Nb Number of blades

xxxi

Ne Number of finite elements

pressure

B Position vector

t Time (sec)

blade frame

wake-fixed frame

xxxii

TPG Transformation matrix from global wake-fixed frame to

preconed frame

frame

frame

frame

blade

u Control vector

on the blade

p

vh Hover induced velocity, CT /2

y State vector

xxxiii

w Vertical speed of the aircraft, flapwise elastic deflection of a point

on the blade

x Trim vector

Greek Symbols

β1c , β1s Tip path plane longitudinal and lateral tilt angles

xxxiv

γxy Coherence

Component of strain

θ0 , θ1c , θ1s , θ0t Collective, lateral, longitudinal cyclic pitch and tail rotor

collective

θBT Pre-twist of swept tip with respect to the junction to the straight

ζ vortex age

κy , κz Beam curvature

xxxv

λt Tail rotor inflow

ν Kinematic viscosity

ρ Air density

σ Stress components

deformation

far wake

angle, rad

( )> Transpose

( )A Aerodynamic

xxxvi

(..)b Straight blade finite element

( )H Horizontal tail

(..)h Harmonics

(..)T Tension

( )V Vertical tail

Abbreviations

BV I Blade-Vortex Interaction

xxxvii

ODE Ordinary Differential Equation

P IP C Pseudo-Implicit Predictor-Corrector

xxxviii

Chapter 1

Introduction

1.1 Motivation

Ever since the dawn of civilization, design and innovation have been a trial-and-error

process. Advancement in technology, whether it was the wheel or the pulley, the

light bulb or the steam engine, occurred as a result of test and failure, and learning

from one’s earlier mistakes. The history of aviation contains many examples of

ingenuity and disaster in the road to the first flight of the Wright Brothers, and for

many years after that inventiveness walked hand in hand with great achievements

but also catastrophic mistakes (Ref. 2). Aviation pioneers often tested their ideas

The development of the helicopter was not free of failures and mistakes. In-

deed, the complexity of rotating-wing aircraft delayed the first successful flight of a

helicopter more than 30 years after the first flight of a fixed-wing aircraft, a period

in which many ideas and configurations were built and tested (Ref. 3). Many of

these ended in crashes, some leading to better understanding of the physics of ro-

torcraft, some discarded. In any case, these early inventors had no means to validate

1

their ideas but to build them and test them, and build them again iterating until a

It is for this reason that having accurate simulation models is vital in the de-

sign process to avoid unnecessary redesigns due to poor predictions (Ref. 4). Such

redesigns not only are expensive, but also delay the overall production process.

Simulation models can be used to validate new technologies before they are built.

Proper understanding of the theory behind rotorcraft flight can therefore help con-

struct better and more accurate simulation models to build better helicopters more

efficiently.

expected to perform missions that require more maneuverability and agility. Such

maneuvers often involve high rates and rapid changes, and accurate simulation re-

quires high resolution models to capture the transient dynamics (Ref. 5). Civil

helicopters are also being put through more demanding mission task elements, such

as steep approaches to land in order to minimize the noise signature over populated

areas, which involve risks such as high vibration or power settling. The availabil-

ity of high fidelity simulation models provides the means to study such phenomena

Design of advanced flight control systems can also greatly benefit from accurate

modeling of helicopters. Flight dynamics simulation can be used for the verification

of new control system designs before they are implemented in real helicopters, spe-

cially if the model can rigorously replicate the real life conditions in a more varied

2

range of flight conditions, without relying extensively on empirical data that might

tion that allow for increasingly more accurate predictions in a broader range of

flight conditions (Ref. 6). While many of such models are too complex and com-

putationally expensive to be used for real time simulation, they have demonstrated

their utility by helping design better and more maneuverable helicopters and reliable

model involves the coupling of many sub-systems, such as the aeroelastic rotor, the

rotor wake, the fuselage, etc, which are not independent of each other and interact

between themselves, adding another level of complexity to the behavior of the heli-

copter model (Ref. 4). These comprehensive models can then be used to calculate

the steady-state trim equilibrium solution under any prescribed flight condition, to

obtain a linearized model that can be used for stability and frequency response cal-

inputs.

dynamics model used to predict the main rotor induced velocities. The existing

flight dynamics simulation models, both in industry and in academia, utilize a wide

spectrum of aerodynamic models to provide the main rotor inflow. Simple inflow

models, such as those based on momentum theory or dynamic inflow, have been

3

widely used in comprehensive flight dynamics simulation models because of their

low computational cost and their relatively easy implementation and solution with

the rest of the rotor-fuselage model. However, they are limited in that they are

based on a rigid and undistorted representation of the rotor wake structure and

cannot capture the dynamics of the wake flow field. Free wake models overcome

such limitations. By modeling the wake vortices as they are released from the blade

and allowing them to distort freely under the influence of their mutual interactions

as well as the blade circulation and the free stream and maneuvering velocities, it is

possible to obtain an accurate description of the flow field and the induced velocities

quest to accurately describe the behavior of helicopters for flight dynamics applica-

tions (Ref. 8). Modeling the non-linear coupling in flap, lag and torsion of the rotor

model developed over the years at the University of Maryland (Ref. 6), commonly

referred to as HeliUM, in two specific areas, namely the modeling of the rotor wake

and the aeroelastic blade model. Regarding the rotor wake, this study will couple

the rest of the rotor-fuselage model. This free wake model can capture maneuvering

flight accurately in time, and therefore can be used to determine the response of

the helicopter to transient pilot inputs. It can also be used for the calculation of

equilibrium flight conditions, both periodic (straight and level flight, coordinated

4

turns) and aperiodic (steep descents). Since most modern helicopters have swept

tips, as they reduce noise and vibrations and improve the overall performance of

rotors (Ref. 8), it is important to include tip sweep in the aeroelastic modeling of

the blade as well, therefore it is part of the present study to expand the blade model

received considerable attention through the years, most often it has been limited to

straight and level flight. Few studies have attempted to model steady coordinated

turns (Refs. 6, 10, 11), or steady climbs and descents (Refs. 10, 11), of which only

Ref. 6, the preceding work to the present study, includes the effect of maneuver-

ing on the free wake (Ref. 12). Most comprehensive flight dynamics models have

the capability to calculate the transient response of the helicopter to arbitrary ma-

neuvers. Unsteady turning flight has often been considered, although traditional

comprehensive codes did not have the capability to include the effect of maneuver-

ing in their modeling of the wake. The ability to capture maneuvering flight, not

limited to small excursions from trim, expanding the range of flight conditions that

can be modeled with a flight dynamics simulation model, must therefore be the goal

descending flight conditions, in which the rotor wake is unsteady and aperiodic.

Whether it is in shallow descents, where the increased closeness of the rotor vortices

makes their interaction more pronounced, or in steep descents, where the vortices

cross over the rotor plane, modeling descending flight is a challenging problem.

5

Several of the most interesting problems that face aerodynamicists and flight dy-

namicists, such as BVI, autorotation modeling and the VRS, fall within this flight

When the rotor descents into its own wake, the increased proximity of the

vortices creates a strongly unsteady flow close to the rotor. As the rate of descent

increases, the upward flow pushes the vortices through the rotor, which enters the

vorticity in the form of a ring at the tip of the rotor. The VRS manifests in strong

thrust fluctuations, an increase in the rotor torque, highly unsteady blade airloads,

aperiodic blade flapping and loss of control, which makes it a significantly dangerous

flight condition when the helicopter is near the ground. The vortex ring state is not a

condition exclusive of axial flight. In steep descents at low speed the VRS manifests

required to continue to increase the rate of descent (Refs. 4, 13). As the rate of

descent increases, the power first decreases. Eventually, the power required to fly

at the same speed reaches a minimum and any increase in rate of descent requires

more power. When power settling occurs, the helicopter continues to descend even

after applying additional power. Since recovery from power settling usually involves

a significant loss in altitude, knowing the boundaries for this condition becomes

The vortex ring state is characterized by largely unsteady flow. In the incip-

ient stages of VRS, the degree of unsteadiness is mild and therefore the conditions

encountered by the rotor blades over successive revolutions do not change consid-

6

erably and a trim solution is possible and represents the real flight conditions well.

However, as the rotor enters into deep VRS, the flow is highly unsteady and the

flight conditions encountered by the rotor blades over successive revolutions, as well

as the different blades in the same revolution, are considerably different. Under

valid and cannot be applied. The proper approach under these flight conditions is

Another important standing problem that still draws a lot of attention is the

prediction of the off-axis response. By off-axis, one refers to the response in a differ-

ent axis to that related to the control perturbed, i.e., the pitch response to lateral

cyclic or the roll response to a longitudinal cyclic. For years, the helicopter com-

munity was at loss at why many simulation models did not capture the direction of

the off-axis response correctly. It was not until the nineties that some explanation

was found. Rosen and Isser (Ref. 14) demonstrated that the distortion of the wake,

up until then ignored, played an important role in the accuracy of the prediction

of the off-axis response. Since then, the importance of the effect of maneuvering

on accurately predicting both on-axis and off-axis responses has lead to many stud-

ies to account for maneuvering effects in the calculation of the main rotor inflow

distribution.

The main objective of the work in this dissertation is to develop a flight dy-

namics model that can be used to calculate equilibrium and transient conditions in

steady and unsteady maneuvers, including steep and axial descents. The present

7

study starts with the model developed by Theodore (Ref. 6), which is a flight dy-

namics model that consists of a coupled rotor-fuselage model, with flexible blades

modeled with a finite element analysis, and the ability to calculate the induced

velocities on the rotor both with dynamic inflow or with the Bagai–Leishman re-

laxation maneuvering free wake model (Ref. 12). The present study will expand

free wake (Ref. 15), and incorporate the swept tip into the modeling of the flexible

blade. The new model will then be used to simulate both steady and unsteady flight

This section presents a review of the current state of the art of those topics relevant

to the present study. It is divided in the following sections: first, a review of existing

of the different methods for determining the rotor inflow, with particular attention

to free wake models, specially those which can capture maneuvering flight. Next,

the problem of the off-axis response in maneuvering flight is covered. The state of

the art in the modeling of descending flight continues, with special emphasis in the

vortex ring state. Finally, a review of the evolution of aeroelastic models with swept

tips is included.

8

1.2.1 Helicopter simulation models

In recent years, several comprehensive flight dynamics simulation models have been

models have achieved high levels of sophistication, allowing for very accurate pre-

namics codes and their inflow modeling characteristics, and a summary is presented

here, together with some updates and newer models that have been made public

since then. A point made by Theodore and worth reinforcing is the difficulty to ob-

tain details of the methodology or solution techniques used by these models, which

are not often made available, particularly with the commercial models. Each of

The CAMRAD family of models (Ref. 11), which include CAMRAD, CAM-

RAD/JA and CAMRAD II has been developed by Johnson Aeronautics. Like all the

fuselage model is rigid, but the blades are modeled with a finite element analysis

that allows for both rigid and flexible blade models with coupled flap, lag and torsion

degrees of freedom. The induced velocities can be calculated with simple uniform

inflow based on momentum theory or with the Johnson free wake model (Ref. 16),

which is a modification of the Scully wake model (Ref. 17). The model can be used

to obtain a trimmed solution, both as free-flight and wind tunnel trim, and is not

limited to level flight but includes coordinated turns and climbing and descending

9

flight. Moreover, the model can be used to obtain a linearized model using a per-

turbation analysis from a trim condition and to obtain the transient response to

pilot inputs, both with the linear and the non-linear models. CAMRAD has been

The CHARM model (Refs. 18, 19), developed at Continuum Dynamics Inc.

(CDI), is a general rotor/airframe model that includes the Constant Vorticity Con-

tour wake model combined with a fuselage aerodynamic model using fast vortex and

fast panel methods. The result is a model that can be used in real time while keeping

the advantages of free vortex models, and that can capture rotor-wake-fuselage in-

teractions. The blades are modeled with a finite element analysis that includes flap

and lag bending, twist and extension. This model has been coupled with Sikorsky’s

GenHel flight dynamics simulation model providing a platform to perform trim and

The COPTER model at Bell (Ref. 20), which stands for Comprehensive Pro-

gram for Theoretical Evaluation of Rotorcraft, includes both a rigid and a NASTRAN-

based flexible representation of the fuselage. The blades are modeled with fully

coupled flap, lag and torsion, and the induced velocities are obtained with the

Scully (Ref. 17) free wake model. It can be used for both free-flight and wind-tunnel

to determine the response to pilot inputs. However, not much validation has been

published.

simulation model. It contains a rigid fuselage model and the flexible blade includes

10

coupled flap and lag, but uncoupled torsional degree of freedom. To calculate the

induced velocities, FLIGHTLAB uses a three state version of the Peters and He

Finite State Dynamic inflow model (Ref. 22). Like most other models, it can be

used to obtain trim, linear models and transient response calculation, being the

ERA and DLR, have joined forces in the development of their own comprehensive

model for flight dynamics, the Helicopter Overall Simulation Tool, or HOST (Ref. 23).

The HOST model can be used to do trim calculations, to perform time domain and

inverse simulations, and to obtain an equivalent linear system. Both a rigid and an

elastic model can be used for the blades. The inflow can be calculated with several

simple models, with the Pitt–Peters dynamic inflow (Ref. 24), although a model to

account for the wake distortions has been included (Ref. 25), with a vortex rings

(Ref. 26) has evolved and been renamed as the Rotorcraft Comprehensive Analy-

sis System (RCAS) (Ref. 27). RCAS has both a rigid and elastic NASTRAN

modal representation of the fuselage. The blades are modeled with a finite element

analysis and include coupled flap, lag and torsional degrees of freedom. For the

theory, Peters–He finite state wake (Ref. 22), a prescribed wake model, and three

free wake models: the Scully-Johnson wake (Ref. 16), the Bagai–Leishman relax-

ation free wake model (Ref. 28) and the Bhagwat–Leishman time-marching free

11

wake model (Ref. 7). Like most of the comprehensive models, it can calculate trim,

The model developed at Boeing, known as the Technology One program (TECH-

01) (Ref. 29), also includes a flexible fuselage model in addition to the rigid one,

but the flexible blades have only coupled flap and pitch degrees of freedom, while

the lag dynamics are uncoupled. Both the Drees (Ref. 30) and Pitt–Peters (Ref. 24)

dynamic inflow models can be used to determine the induced velocities, but no free

wake models are available. Like with the Bell model, TECH-01 can be used for trim,

linearization and integration calculations, but little validation has been published.

Developed also at the University of Maryland, the UMARC model (Ref. 31)

can use both rigid and flexible fuselage and blades, with fully coupled flap, lag and

torsion. UMARC has several options to calculate the induced velocities, including

simple momentum theory uniform inflow, Pitt–Peters dynamics inflow (Ref. 24), the

Scully (Ref. 17) and Johnson (Ref. 16) wakes and the Bagai–Leishman relaxation

free wake model (Ref. 28). UMARC can be used to calculate an algebraic trim

solution and a linearized model, but can only obtain the response to pilot inputs

over the years, called HeliUM in its current implementation, that derived from the

original GenHel (Ref. 32). This model includes flexible blades with coupled flap,

lag and torsion dynamics, a rigid fuselage with aerodynamic data obtained from

look-up tables, several dynamic inflow options and two free wake models, the Bagai–

Leishman relaxation free wake (Ref. 28) and the Bhagwat–Leishman time-marching

12

model (Ref. 7). The model can obtain trim equilibrium conditions, linearized models

and integrate the equations of motion to obtain the response to arbitrary pilot

in Section 1.3.

model used to predict the main rotor induced velocities. The existing flight dy-

namics simulation models, both in industry and in academics, utilize a varied set of

The simplest inflow models are those based on momentum theory or a combination

of momentum theory with blade element theory (BET) (Ref. 3). Being computa-

tionally inexpensive and reliable, these models are still extensibly used today as a

preliminary analysis tool. Linear models, such as that of Drees and Mangler and

Squire improve the previous methods with variations of the longitudinal and lateral

Dynamic inflow models are some of the most popular and extensively used

inflow models in comprehensive flight dynamics models. They have the general

advantage of being formulated in first order form, which simplifies their integration

13

equations in state-space form with an equal number of states representing the inflow

coefficients. The matrices associated with such formulation represent mass and gain

functions and can be obtained with different methods, from actuator disc to vortex

theory. The different dynamic inflow models available were reviewed by Gaonkar

One of the most widely used dynamic inflow models in the literature is the

Pitt–Peters model (Ref. 24), which derives the inflow equations using unsteady-

actuator disc theory, relating the rotor transient loads to the induced velocity field.

More recently, finite state wake models, like that of Peters and He (Refs. 22,

36), have allowed for a formulation with multiple states which permit both radial

and higher harmonic azimuthal variations of the rotor inflow. A more realistic

approximation to the induced velocity field can be obtained with such a method by

increasing the number of states, while maintaining the formulation in first order.

The basic dynamic inflow models have been extended through the development

of sets of coefficients that allow the application of such models to specific flight

conditions. One example is that of Prasad et al. (Refs. 37, 38), which includes the

dynamic wake distortion effects due to maneuvering, and which obtains the time

constants of the model using a vortex tube analysis. Another extension is that of

He et al. to allow the finite state wake model to be used for such problems as the

prediction of the off-axis response or flight through the vortex ring state (Ref. 39).

As it has been mentioned already, dynamic inflow models benefit from their

formulation being in a form that is easy to couple with comprehensive flight dynam-

ics codes and from their inexpensive cost of computation. However, there are some

14

practical disadvantages associated with the use of such models, particularly the fact

that they are defined with a set of coefficients associated with a particular flight

Moreover, dynamic inflow models are based on a rigid and undistorted representa-

tion of the wake structure, and cannot account for the dynamics of the wake flow

field.

Vortex models describe the rotor wake as a series of vortex filaments released from

the blade and that are discretized in time and space, and once the position of each

models, or assumed rigid, as in prescribed wakes), the induced velocities on the rotor

can be calculated from the influence of each vortex segment in the flow field (Ref. 3).

The studies of Bhagwat (Ref. 15) and Ananthan (Ref. 40) contain a detailed review

of the different types of vortex models of the rotor wakes and the different attempts

that have been made at modeling them. Therefore, only a summary is included

here.

Prescribed wake models are the simplest of all the vortex wake models. The

geometry of the wake is described with undistorted helical vortex filaments, the

influence of which on the rotor can be calculated with Biot-Savart law to obtain the

induced velocities. Prescribed wake models have been used successfully to model

both hovering (Landgrebe (Ref. 41), Kocurek & Tangler (Ref. 42)) and forward

15

flight (Egolf & Landgrebe (Ref. 43), Beddoes (Ref. 44)), but they need to rely on

empirical data to model the distortions of the wake structure. For this reason, and

because the wake is assumed rigid, prescribed wake models provide little advantage

Free vortex models represent the next level of sophistication in the calculation

of the induced velocities. Unlike prescribed wake models, free wake models include

the geometry and strength of the vortex filaments as part of the problem. The vortex

filaments are allowed to distort freely under their self and mutual influence, including

the influence of the blade bound vortex and the velocities due to translation and

Free wake models can be categorized according to the method used to discretize

the flow field or the numerical scheme used to solve them. The vortex filaments are

usually approximated with straight line segments, although they can also be modeled

with curved vortex segments (Ref. 45) or vortex blobs (Ref. 46). Regarding the

solution scheme used, the models are either relaxation or time-marching methods.

Relaxation methods assume that the wake solution is periodic and obtain

a solution through an iterative process until the wake geometry does not change

and free of the instabilities that are associated with time-marching schemes, they

The first attempt at using an iterative method to model the rotor wake was

done by Clark & Leiper (Ref. 47), who modeled two turns of the tip vortex and used

ring vortices for the far wake and obtained successful results for hover performance.

16

The wake model of Scully (Ref. 48), which he developed as a relaxation scheme

one of the most recognized and widely used today. The induced velocities were

calculated with a weighted average using a relaxation parameter, which gives name

to this family of models. It used an azimuthal step of 15 degrees, and was used

successfully both in hover, with 6 to 12 free wake turns and a vortex cylinder for

the far wake, and in forward flight, for which 2 to 4 free wake turns were deemed

sufficient. Its success is traced to the fact that it was later adapted by Johnson and

model. Miller (Ref. 50) developed a model for hover with vortex sheets or a vortex

cylinder to model the far wake boundary, and included the Weissinger-L lifting

surface model (Ref. 51) to calculate the bound circulation. Bliss et al. followed with

a family of relaxation models (Refs. 52, 53) characterized by curved vortex segments,

large core radius (5% of the rotor radius). Miller & Bliss (Ref. 54) developed an

interative scheme which differed in that the non-linear equations of the wake were

Closer to the present work is the wake model of Crouse & Leishman (Ref. 55),

which precedes the Bagai–Leishman free wake model (Ref. 28). The Bagai–Leishman

wake model uses a pseudo-implicit predictor-corrector scheme to solve the wake ge-

ometry, a five point central differencing scheme to approximate the derivatives both

in time and space and a Weissinger-L model to solve the blade bound circulation-

lift problem. This wake model, capable of capturing the effects of maneuvering,

17

was coupled successfully with several flight dynamics models, including that of

Time-marching free wake models do not enforce periodicity of the rotor wake

time and allowing the wake to distort freely until a solution is found.

Several of the earliest attempts at solving the wake problem with time-marching

method were unsuccessful due to their choice of numerical scheme leading to nu-

merical instabilities (Refs. 17, 56, 57). Landgrebe (Ref. 58) overcame some of the

a coarse discretization to reduce the high cost of the method. He obtained good

correlation with experimental data for forward flight, but in hover he could not

overcome the numerical instabilities. Similarly, Sadler (Ref. 59) used an explicit

Euler scheme for multiple rotor configurations, which worked well in forward flight

but was too unstable in hovering flight. A method with curved filaments and a

predictor corrector scheme was used by Bliss et al. (Refs. 60, 61), but only used for

forward flight conditions. Also for forward flight use only was the time-marching

wake model of Egolf (Ref. 62), which used a vortex lattice to represent the wake.

The CHARM wake model (Refs. 19, 45) uses a constant vorticity contour

(CVC) method, which follows lines of constant strength vortex elements, with curved

vortex filaments.A vortex lattice lifting surface model is used for the blade, which

uses a finite element structural analysis to determine the mode shapes. The fuse-

lage body and blade surface are modeled with panel methods, and can therefore

be used to study the interaction between the rotor wake and the helicopter body.

18

Fast computational methods and a reduced wake resolution are used to reduce the

A different approach has been taken more recently by Brown et al. (Refs. 63,

64), which use a vorticity transport method to solve the wake, which is solved in an

Eulerian frame using Toro’s Weighted Average Flux algorithm (Ref. 65) to march

the equation in time, which provides the advantage over other CFD methods of

maintaining more control over the dissipation of vorticity in the wake. The model

marching free wake model developed from the Bagai-Leishman model (Ref. 12). This

is the model used in the present study. The Bhagwat model uses a second-order,

of the wake in time, while the space derivative is approximated with a five-point

central difference scheme. This model has been tested and validated over many flight

conditions. In addition, Ananthan (Ref. 40) has added vortex filament stretching

and has demonstrated the validity of the model in unsteady maneuvers (Ref. 5).

els has increased rapidly in the last decade, especially for nonreal-time, research

type simulation models. One of the drivers in the research community has been the

problem of predicting the off-axis response (e.g., the pitch response to lateral cyclic).

19

The problem remained unresolved until Rosen and Isser (Ref. 14) demonstrated the

role played by the distortion of the wake geometry, due to the maneuver, using a

prescribed wake approach. Since then, several approaches have been followed to

account for maneuvering effects in the calculation of the main rotor inflow distribu-

tion, though most use a dynamic inflow type model. Basset and Tchen-Fo (Ref. 25)

developed a set of coefficients using a dynamics vortex model to couple the angular

rates to the inflow distribution, and used them in a dynamic inflow model. The

popular Pitt–Peters dynamic inflow model has also been extended to capture the

dynamic distortion of the wake due to maneuver and transition flight. The work by

Prasad, Peters et al. (Refs. 37, 38), where the appropriate time constants are ex-

tracted from a vortex tube analysis, is one of the latest developments of the model.

Simple but accurate state-space inflow models suitable for maneuver analyses have

also been extracted using frequency domain system identification, either from ex-

A free vortex wake model to account for the effects of maneuvering from first

principles, has been developed by Bagai and Leishman (Ref. 69). Note that, in the

context of the present paper, “from first principles” is limited to meaning that no

a priori assumptions are made on the distortion of the wake geometry caused by

the maneuver, that is, the wake geometry is free to evolve from the time history of

the motion of the rotor blades, and that no empirical or semi-empirical correction

coefficients are used in the definition of the rotor inflow or wake geometry. The model

of Ref. 69 is based on the solution of the vorticity transport equations, and uses a

20

earlier). Therefore, it is rigorously appropriate only for a steady trimmed flight

condition, and not for transient conditions. The limitations of the relaxation wake

have been subsequently removed in the time-accurate free wake model of Bhagwat

and Leishman (Ref. 9), which is therefore suitable for analyzing unsteady maneuvers

of arbitrary amplitude.

Another free wake model capable of modeling maneuvering flight has been

developed by Wachspress et al. (Ref. 19) as part of the simulation model CHARM

(mentioned earlier), and successfully applied to the analysis of flight in vortex ring

state and to the prediction of off-axis response. This wake has been coupled with

the Sikorsky GenHel simulation code, and used for the modeling of free wake-

flight test data (Ref. 70). The wake has also been coupled with the NASA ver-

sion of the GenHel code (Ref. 71), giving results similar to those of the Pitt–Peters

dynamic inflow model. The same simulation code has also been used to obtain air-

craft and blade motion data for wake dynamics and acoustics studies (Refs. 72, 73).

In this case, aircraft and blade motion data were provided as input to the time-

accurate free wake model of Ref. 9 in an “open loop” fashion, as the NASA GenHel

model is based on linear inflow, and rigid flap and lag motion with an empirical

The relaxation wake of Ref. 69 was coupled with a flight dynamic simulation

that included a flexible blade model, and the resulting model was used to study

the dynamics of the Eurocopter BO-105, with a special focus on the prediction of

the off-axis response (Ref. 74). One of the conclusions of the study was that an

21

accurate prediction of the off-axis response from first principles was indeed possible,

but it required the simultaneous modeling of: (i) blade flexibility in flap, lag, and

torsion, and (ii) distortion of the wake geometry caused by the maneuver. The

off-axis predictions with the correct wake distortions but with elastic flap only, and

those with fully coupled flap-lag-torsion but with a conventional free wake without

maneuver distortion, exhibited the traditional “wrong direction” error (Ref. 74). On

the other hand, accurate predictions of the on-axis response did not require such

a sophisticated modeling. Although the relaxation free wake was not rigorously

Experimental investigations of the VRS started in the 1950s with the wind-tunnel

work of Drees & Hendal (Ref. 1) who, via smoke-flow visualization, showed the

recirculation of the flow in the rotor plane and the toroidal shape of the vortex

ring around the tips of the rotor blade (Fig. 1.1). Refs. 75 and 76 provide a fairly

Numerical analysis of the VRS has not been very abundant. It was necessary

that the technology matured and provided the capability to accurately model the

vortex flow field. Simple models, such as those based on momentum theory, can

be used when the descent angles are not very large. Several early works, such as

those in Refs. 77 and 78, used momentum theory to map the boundaries of the

22

vortex ring state. The model in Ref. 13 explains power settling and autorotation,

and provides contour maps for the power required by the rotor as a function of

glide speed, glide angle and rotor pitch angle. More recently, several studies have

proposed simple inflow models (Refs. 39, 76, 79–82) to work in steep or axial descent

theory can be adequate with the appropriate selection of tuning parameters, however

they are not based on first principles and are unable to capture the dynamics of the

flow field of the rotor wake. The most accurate numerical approaches to model

the VRS are those based on free wake models and computational fluid dynamics.

Leishman et al. (Refs. 83, 84) studied the instabilities associated with the onset of

the VRS using a time-accurate free wake model (Ref. 9). This model can capture

the formation and development of the vortex ring state and can be used both to

in descending flight. Brown et al. (Refs. 75, 85) developed a vorticity transport

model of the rotor wake, which they successfully applied to both axial and forward

flight descents. Their results allowed for a numerical mapping of the VRS boundary

that complements that obtained through experiments and flight test. Moreover,

the interference effects between the rotor wake and the fuselage and tail during the

Not much work has been done to implement such computational models ca-

pable of accurately describing descending flight and VRS into comprehensive flight

dynamics simulation models. Descent flight and autorotation are the drivers behind

the implementation of coupled advanced wake and flight dynamics models such as

23

Ref. 86, which compares the results for finite-state induced inflow methods with the

vorticity transport wake model. Their results show that, while for shallow descents

the differences are small, for steeper descents, such as those required for autorota-

tion, there were some discrepancies between the two models and the higher accuracy

free wake (Ref. 9) is used in Ref. 73 to perform aeroacoustics calculations for pop-up

In the early eighties, several studies investigated the effects of swept blade tips on

rotor aerodynamics (Refs. 87–89). These early studies showed the importance of

proper modeling of the blade tip sweep, which introduces flap-torsion and lag-axial

coupling effects. Moreover, because elastic displacements and dynamic pressure are

the largest at the tip, proper dynamic and aerodynamic modeling becomes necessary

Tarzanin & Vlaminck (Ref. 90) conducted the first analytical study of the

effect of swept tips on hub loads. Their approach to model the swept tip consisted

of displacing the shear, aerodynamic and mass center of cross sections on the tip.

They concluded that tip sweep affects both vibration and stability.

The first systematic aeroelastic model of rotor blades with swept tips was

24

developed by Celi & Friedmann (Refs. 91–93). This model was based on the for-

use a finite element approach based on the Galerkin Method of Weighted Residuals,

and a special finite element model for the tip was developed. The main conclu-

sions of their study are that the flap-torsion and lag-axial couplings introduced by

the tip sweep may produce aeroelastic instabilities related to frequency coalescence,

but when frequency coalescence does not occur, tip sweep has a stabilizing effect.

This model’s main limitations were the assumption that the transformation at the

joint between the straight and swept part of the blade is linear and that only tip

parameters.

The nonlinear nature of the transformation at the joint between the tip element

and the straight part of the blade was derived by Panda in Ref. 97. For sweep angles

greater than 20 degrees, he demonstrated that the nonlinearities are significant and

Benquet & Chopra (Ref. 98) developed an aeroelastic analysis of rotor blades

with both sweep and anhedral using a finite element method based on Hamil-

ton’s principle, to calculate the loads and response in forward flight. Kim &

Chopra (Ref. 99) developed the model of Ref. 98 further by including the non-

linear transformation of Ref. 97 at the junction between the straight blade and the

tip. In addition to this, Bir & Chopra (Ref. 100) developed the model for advance

Some more recent studies by Yuan & Friedmann (Refs. 101, 102) considered the

25

aeroelastic behavior of composite blades with swept tips and examined the structural

couplings of such blades. The blade model included arbitrary cross-sectional shape,

A structural optimization study was performed, with the composite ply orientations,

tip sweep and anhedral as the design variables, with the objective of minimizing

vibratory hub loads in forward flight. Posterior studies (Ref. 103) combined this

The origin of the model used in the present study is traced back to the GENHEL

helicopter model developed by Howlett in Ref. 32. The rotor in this model was

limited to rigid blades hinged in both flap and lag. Torsion was accounted for by

means of an empirically-based dynamic twist model. A rigid body model was used

for the fuselage, with its aerodynamics obtained from look-up tables containing

empennage and tail rotor in the present study are based mostly in this original

model.

Ballin (Ref. 104) extended the model, his primary contribution being the en-

gine modeling. Further improvement to the model, now called UM-GENHEL, was

carried out by Kim (Ref. 105), who can be credited with two important contribu-

tions. First of all, he added the Pitt–Peters dynamic inflow model (Refs. 24, 106)

to calculate the main rotor inflow. In addition, he reformulated the trim process

26

to represent the equations of motion in rigorous first-order form, which allowed the

calculation of linear, constant coefficient equations of motion that describe the small

The modeling of the rotor was improved by Turnour (Ref. 107), who included

a fully numerical structural formulation (Ref. 108) with coupled elastic flap, lag and

torsional degrees of freedom, plus a finite element analysis and a modal coordinate

also improved the aerodynamic modeling with a new higher order dynamic inflow

model (Refs. 22, 36), which includes the effects of trailed as well as shed wake, and

This model, referred to as FLEXUM, was specialized for the UH-60 helicopter and

could be used to calculate trim, linearize the equations of motion as well as time

Theodore (Ref. 6) made a significant contribution to the way the model, that

he termed HeliUM, computed the inflow, with the addition of the maneuvering free

wake model of Bagai and Leishman (Ref. 110). In addition, he included Keller’s ex-

tended momentum theory for inflow model and a quasi-steady aerodynamic model

for blade section aerodynamics. The model, which allowed for more accurate pre-

dictions of the off-axis response, was also modified for a hingeless rotor, and a full

Besides an optimization and sensitivities study, the work of Fusato (Ref. 111)

provided a new implicit formulation of the equations of motion, which makes the

simulation model more flexible and modular. The last improvements to the model

27

before the present work were carried on by Cheng (Ref. 112), who modified and

applied the model for the study of Higher Harmonic Control, extracted a high-order

linear model including the HHC characteristics in rotor states and hub loads and

extended the vibration prediction capabilities. Presently, the model evolves through

the work of Malpica (Ref. 113), who has performed a closed-loop HHC and time

periodic stability analysis and has incorporated a trailing-edge flap model for flight

and rotor control analysis and a state-space formulation for compressible unsteady

aerodynamic forces.

maneuvering free vortex model, the Bhagwat–Leishman, wake model, that can

arbitrary amplitude. A new approach for the trim and integration calcula-

calculate the bound circulation to the rest of the flight dynamics model.

2. To improve the fidelity of the aeroelastic blade model with the inclusion of

the modeling of the swept tip. The implementation of the swept tip has been

a model are not really a specific objective of the present study. However,

28

the effect of including the blade with tip sweep on the trim solution and the

4. To use the developed model to study the physical mechanisms behind steady

and transient helicopter flight. In particular, steady level flight, turning flight

and descending flight equilibrium conditions will explored, and the response

5. To validate the proposed mathematical model with flight test data when avail-

a time-accurate free wake model to the trim solution and to the integration of

trim and to integrate the response of the helicopter to pilot inputs accurately

in time.

29

2. Optimized the trim procedure to reduce the overall cost of computation. Un-

necessary calls to the free wake model were eliminated, and the circulation-

inflow dependance was solved in a more efficient manner than in the previous

3. Introduced the modeling of the swept tip into the aeroelastic blade model.

This adds one more level of sophistication to the modeling of the blade, which

a new approach that makes use of the Weissinger-L model available with the

free wake model, which not only provides a more accurate value of the vortex

strength, but also allows the inclusion of the near wake model as well as permits

a frequent series of updates of the bound vortex with every new value of the

is presented. This description is not complete, but illustrates the model and its

capabilities.

Chapter 3 describes the free wake model used for this study in detail.

Chapter 4, discusses the different solution methods used in the model, namely

the calculation of the blade mode shapes, the trim solution and the determination

30

of the response of the helicopter to pilot inputs. In addition, the coupling of the

wake model with the trim and integration methodologies is discussed, and a solution

offered.

this dissertation is described, and the natural frequencies and blade mode shapes

are shown.

In Chapter 6, results are presented for the calculation of trim with the new

additions to the model. These are compared to experimental data. Results are

obtained for level flight, turning flight, descending flight and descending turning

flight.

In Chapter 7, the results of applying the model for integrating the equations of

motion are described. The response is obtained for different maneuvers at different

speeds, and the model is compared to flight test data when available.

31

Figure 1.1: Smoke-flow visualization of the vortex ring state (Ref. 1).

32

Chapter 2

Mathematical Model

This chapter contains a description of the mathematical model of the helicopter used

in this study. The chapter starts with an overview of the main features of the model,

followed by a description of the aircraft, rotor and free wake coordinate systems.

Next, the formulation of the main rotor equations of motion and the rigid fuselage,

the inflow models available in the model, particularly the free wake model used in

the present study. The transformation of the equations of motion into rigorous first

order, state space form is presented next. Finally, a list of the main assumptions

2.1 Overview

Except for the free wake, the mathematical model is formulated as a system of first

ẏ = g(y, u; t) (2.1)

33

or of the equivalent implicit form:

f (ẏ, y, u; t) = 0 (2.2)

equations, which result from the discretization in space and time of a vector partial

differential equation (PDE), the vorticity transport equation (Ref. 15). These finite

from the main flight dynamics model because of the different mathematical form.

Therefore, the wake model does not explicitly contribute any additional states to

the basic flight dynamics model, nor does it change the basic first order form of the

equations of motion.

When the inflow distribution over the main rotor is calculated using the free

wake model, the state vector y takes the form (for a four-bladed rotor):

y(ψi ) = [uF vF wF p q r φF θF ψF λt q11 q21 q31 q41 q̇11 q̇21 q̇31 q̇41

where uF , vF , wF , p, q, and r are the velocities and rates in the body fixed coordinate

system; φF , θF , and ψF are the Euler angles of the fuselage; λt is the tail rotor inflow;

and qik and q̇ik are the generalized displacement and velocity coordinates for the i-th

blade and the k-th normal mode in the rotating frame at the azimuth angle ψi .

additional states are used to describe the main rotor inflow, and the state vector y

34

becomes:

y(ψi ) = [uF vF wF p q r φF θF ψF λ0 λc λs λt q11 q21 q31 q41 q̇11 q̇21 q̇31 q̇41

where λ0 , λc , and λs are the uniform, cosine and sine coefficients of the dynamic

where θ0 and θt are the main and tail rotor collective controls, θ1c and θ1s are the

lateral and longitudinal cyclic controls and θ̇0 , θ̇1c and θ̇1s are the time derivatives

of the pitch controls, which are used only in a few occasions in the model, such as

Several different coordinate systems are used throughout the present study, for the

body as a whole, for the rotor and for the free wake. This section defines the different

The overall motion of the helicopter is described in three main body coordinate

systems: the inertial coordinate system, the body-fixed coordinate system and the

35

wind coordinate system. All three frames of reference have their origin at the center

The inertial coordinate system has its origin at the helicopter CG, and its zI axis

is vertical, and therefore it has the same direction as the weight vector. The xI and

yI axes can have an arbitrary direction. They are usually taken pointing North and

East, respectively, but the precise orientation is not important for the present study.

The body-fixed coordinate system, shown in Fig. 2.1, is also centered at the CG, and

its axes rotate with the fuselage. The xB and zB axes point forward and downward

respectively, and define the vertical plane of symmetry (or quasi-symmetry) of the

is based on the Euler angles φ, θ and ψ, which represent the roll, pitch and yaw

attitudes of the aircraft. The sequence and rotation of these angles is shown in

36

Fig. 2.2. The resultant transformation matrix is:

cos θ cos ψ sin ψ cos θ − sin θ

sin φ sin θ cos ψ sin φ sin θ sin ψ

sin φ cos θ

[TBI ] = − cos φ sin ψ (2.6)

+ cos φ cos ψ

cos φ sin θ cos ψ cos φ sin θ sin ψ

cos φ cos θ

sin φ sin ψ − sin φ cos ψ

Therefore:

i i

B I

jB = [TBI ] jI (2.7)

k

k

B I

The wind coordinate system has its center at the CG and its orientation is deter-

mined by the freestream airflow. The xW axis points forward, aligned with the

the right and downwards respectively. The relation between the wind axes and the

body-fixed frame is determined by the fuselage angle of attack αF and the fuselage

sideslip angle βF , and it is shown in Fig. 2.3. The transformation between both

cos αF cos βF − cos αF sin βF − sin αF

[TWB ] = (2.8)

sin βF cos βF 0

sin αF cos βF − sin αF sin βF cos αF

37

Thus:

i i

W B

jW = [TWB ] jB (2.9)

k

k

W B

In the present study, this coordinate system is used only to obtain the aerody-

namic characteristics of fuselage and empennage, which are defined in tabular form

The rotor shaft is not usually aligned with the zB body axis, but tilted a small

amount to help alleviate the steady loads it sustains in flight. The shaft coordinate

system, which has its origin at the hub and is a body-fixed axis system, has its

zS component in the shaft direction, the xS axis pointing forward, and the yS axis

pointing to starboard.

The angles iθ and iφ are, respectively, the lateral and longitudinal tilt angles

of the shaft. The transformation matrix between the shaft and the body-fixed

cos iθ sin iθ sin iφ sin iθ cos iφ

[TSB ] = − sin iφ (2.10)

0 cos iφ

− sin iθ cos iθ sin iφ cos iθ cos iφ

so that:

i i

S B

jS = [TSB ] jB (2.11)

k

k

S B

38

Hub rotating coordinate system

The hub rotating coordinate system has its origin at in the hub, with its z axis is

aligned with the shaft, and it rotates about the shaft at the rotor speed. The z axis

points upwards, the x axis lies on a vertical plane containing the elastic axis of the

undeformed blade and is positive outwards, and the y axis is perpendicular to them

and positive in the forward (i.e., lead) direction. The transformation between the

shaft fixed and the hub rotating frames is illustrated in Fig. 2.4, and determined by

− cos ψ sin ψ 0

[TRS ] = sin ψ cos ψ 0 (2.12)

0 0 −1

Thus:

i

i

S

j = [TRS ] jS (2.13)

k

k

S

The undeformed preconed blade coordinate system is the result of rotating the hub

rotating coordinate system by the blade precone angle βp . This rotating frame has

its center at the blade root; the êx unit vector is positive outwards, the êy vector is

oriented in the blade lead direction and the êz is perpendicular to the undeformed

elastic axis and is positive upwards. Figure 2.5 shows the relation between the hub

39

The transformation from the hub rotating frame and the blade undeformed

cos βp 0 sin βp

[TPR ] = (2.14)

0 1 0

− sin βp 0 cos βp

so that:

ê i

x

êy = [TPR ] j (2.15)

ê k

z

The deformed blade coordinate system is a rotating coordinate system whose origin

is at the point on the deformed elastic axis of the blade corresponding to the cross

section being analyzed. The ê0x unit vector points outboard along the tangent to the

elastic axis at the origin, the ê0y axis is aligned with the blade chord perpendicular to

the elastic axis and is positive in the blade lead direction and the ê0z is completes the

orthonormal trio and is defined as positive up. This coordinate system is depicted

in Fig. 2.5.

The transformation from the undeformed preconed to the deformed blade co-

S S S13

11 12

[TDP ] = S21 S22 (2.16)

S23

S31 S32 S33

40

where

S13 = − sin θy

and where,

θx = φ

w,x

sin θy = − p

1 + 2u,x + u,x 2 + v,x 2 + w,x 2

p

1 + 2u,x + u,x 2 + v,x 2

cos θy = p

1 + 2u,x + u,x 2 + v,x 2 + w,x 2

v,x

sin θz = p

1 + 2u,x + u,x 2 + v,x 2 + w,x 2

1 + u,x

cos θz = p

1 + 2u,x + u,x 2 + v,x 2 + w,x 2

After substituting all the terms and simplifying, the transformation matrix

becomes:

1 v,x w,x

[TDP ] = −v,x − φw,x 1 − φv,x w,x (2.17)

φ

−w,x − φv,x −φ − v,x w + , x 1

41

The deformed and undeformed frames are related by the following equation:

0

ê ê

x x

0 = [T ] (2.18)

êy DP êy

ê0

ê

z z

The tip path plane coordinate system is required for the formulation of the dynamic

inflow model. This system is a nonrotating frame, and its zT P P axis is perpendicular

to the plane defined by the first harmonic of the flapping motion of the blade tips,

The transformation between the shaft and the tip path plane coordinate sys-

cos β1c sin β1c sin β1s sin β1c cos β1s

[TTS ] = − sin β1s (2.19)

0 cos β1s

− sin β1c cos β1c sin β1s cos β1c cos β1s

Nb

2 X wtipj

β1c = cos ψj (2.20)

Nb j=1 R − e

Nb

2 X wtipj

β1s = sin ψj (2.21)

Nb j=1 R − e

The relation between the unit vectors of the two involved systems is:

i i

TP P S

jTP P = [TTS ] jS (2.22)

k

k

TP P S

42

Blade sectional aerodynamics coordinate system

The blade sectional aerodynamics coordinate system is needed for the calculation of

the section aerodynamic loads at any point along the blade. Its origin is at the point

of the elastic axis corresponding to the cross section being analyzed. Its unit vectors

determining are defined such that eT points in the lag direction, eR is tangent to the

elastic axis and points outwards and eP is perpendicular to both and is considered

VA = UT eT + UP eP + UR eR (2.24)

sin ζ − cos ζ 0

[TAP ] = sin β cos ζ sin β sin ζ − cos β (2.25)

cos β cos ζ cos β sin ζ sin β

where β and ζ are the local flap and lag slopes of the blade elastic axis with

reference to the undeformed preconed blade coordinate system and are given by:

β = w,x (2.26)

ζ = v,x (2.27)

where w and v are the flap and lag displacements of the elastic axis.

43

The relationship between the two systems is therefore:

e ê

T x

eP = [TAP ] êy (2.28)

e

ê

R z

If the swept tip is being included in the modeling of the blade, additional coordinate

systems are required to formulate the aerodynamic forces and moments locally at

the tip.

The undeformed tip coordinate system has its origin at the junction between

the straight portion of the blade and the swept tip, as shown in Fig. 2.6. The iH axis

points outwards of the swept tip, the jH axis is parallel to the chord of the tip portion

of the blade and perpendicular to the elastic axis, and the kH is perpendicular to

The conversion between the blade and the tip undeformed coordinate systems

is:

i ê

H x

jH = [TΛ ] êy (2.29)

k

ê

H z

cos Λ − sin Λ 0

[TΛ ] = sin Λ cos Λ 0 (2.30)

0 0 1

44

The transformation between the vectors of nodal degrees of freedom, however,

Section 4.2.1.

The origin of the deformed tip coordinate system is at the point of the elastic axis

of the swept part of the blade corresponding to the section being considered. The

iT axis of this coordinate system, shown in Fig. 2.6, is tangent to the elastic axis

at the origin, with the jT parallel to the chord and perpendicular to the elastic axis

and and positive forward and the kT orthonormal to both and positive upwards.

The relation between the deformed and the undeformed tip coordinate systems

is:

i i

T H

jT = [TDP ]L jH (2.31)

k

k

T H

where the subscript L refers to the local tip frame. The transformation matrix

[TDP ]L is then:

1 v,x w,x

[TDP ]L = −v,x − φw,x 1 − φv,x w,x (2.32)

φ

−w,x − φv,x −φ − v,x w + , x 1

45

2.3 Main assumptions

The following list is a summary of the main assumptions that are used in the for-

2. The effects of dynamic stall are not included. Quasi-steady stall and compress-

ibility effects are modeled through tables of lift, drag and moment coefficients.

The unsteady aerodynamic effects include circulatory effects and the accelera-

tion type non-circulatory effects are neglected with the exception of the pitch

damping.

3. Aerodynamic forces and moments on the blade section are based on the airflow

4. The effect of the downwash of the rotor on the fuselage and empennage is not

included.

5. The pitch angle of the horizontal stabilizer (for the UH-60 helicopter used in

this dissertation) is fixed for a given flight condition, and the control logic for

6. The aerodynamic loads on the swept tip are calculated for a cross section

perpendicular to the swept elastic axis, rather than for a sheared airfoil per-

46

7. Independence principle is assumed, such that sweep does not cause a cross

flow.

1. The airframe is a rigid body with a constant mass and the xB -zB plane is a

plane of symmetry.

2. The undeformed blade has no droop or torque offsets. The baseline blade is

3. The blade is rigid in flap, lag and torsion, and generates no aerodynamic loads,

4. The flap, lag and pitch hinges (when they are present) are coincident.

5. The blade cross sections are symmetric with respect to the major principal

axes.

6. The blade cross-sectional area centroid and elastic axes are coincident, which

means that the tension center is coincident with the elastic axis. However,

7. Blade chord, built-in twist, stiffness and mass properties, and cross-sectional

offsets are defined at discrete spanwise stations, and vary linearly in between.

47

9. All blades have identical mass, stiffness, and geometric properties.

10. Bernoulli-Euler beam theory is used, implying that plane cross sections remain

plane and perpendicular to the elastic axis during deformations. The effects

11. The blade undergoes moderate deflections in bending and torsion, implying

12. The blades rotate at a constant angular speed, Ω. Engine and engine control

13. The blade pitch control system, including the actuators, is infinitely stiff.

Freeplay in the control linkages is not modeled. The swashplate and tail rotor

2.4.1 Overview

This section describes the mathematical model of the main rotor used in the present

study. An extended description of the model can be found in Refs. 107, 6. Only a

summary is presented in this dissertation, except for those topics that are of special

relevance for this study, or that have not been included in previous references.

The main rotor blade dynamics are calculated in the rotating frame and sepa-

rately for each blade, allowing the model to be used for dissimilar blades, although

48

the present study will assume identical blades. Each blade is considered a flexible

beam undergoing coupled flap, lag, torsion, and axial motion and is attached to a

hub that may have large amplitude linear and angular motions. The resultant equa-

tions of motion are nonlinear, coupled, partial differential equations with periodic

quires the knowledge of the absolute velocity seen by each point along the blade

elastic axis. This velocity is the derivative with respect to time of the position

vector of a point P on the elastic axis of the blade relative to a fixed point, given

by:

RP = RCG + RH + RB (2.33)

where RCG is the position vector of the body center of gravity with respect to a

fixed point, RH is the position of the hub relative to the center of gravity and RB

is the position vector of the point P , located on the elastic axis of the blade, with

49

respect to the hub.

The position vector of the hub in reference to the center of gravity is given by:

RH = xH iB + yH jB + zH kB (2.34)

The position vector of the point on the elastic axis with respect to the hub is

given by:

where e is the offset of the blade flap, lag and pitch hinges from the axis of rotation

(if present), which are assumed coincident. The elastic portion of the blade starts

at the hinge, and inboard from this point the blade is assumed rigid. The term x0

is the distance from the start of elastic portion to the point P on the elastic axis

of the undeformed section and u, v and w are the elastic deflections of the point

from the undeformed blade coordinate system. Using the coordinate transformation

presented in Eq. (2.14), Eq. (2.35) can be rewritten in the undeformed preconed

coordinate system as

VP = = + + (2.37)

dt dt dt dt

dRCG

= uF iB + vF jB + wF kB (2.38)

dt

50

where uF , vF and wF are the linear velocity components in the body-fixed axes

system.

dRH ∂RH

= + ω × RH (2.39)

dt ∂t

dRB ∂RB

= + ω × RB (2.40)

dt ∂t

in which p, q and r are the roll, pitch and yaw rates, respectively.

The term ∂RH /∂t in Eq. (2.37) is zero because the fuselage is rigid. The partial

derivative ∂RB /∂t is the velocity vector of the point as seen in the body-fixed axes,

and it is

∂RB ∂RB

= + Ω × RB (2.42)

∂t ∂t R

where the first term represents the velocity vector of the point relative to the hub

in the rotating frame and Ω is the angular velocity vector of the main rotor

Ω = ΩkS (2.43)

dRCG ∂RB

VP = + + Ω × RB + ω × [RH + RB ] (2.44)

dt ∂t R

The total local velocity at the blade section is obtain by adding the contribu-

VT = VP − VI (2.45)

51

where the induced velocity VI is subtracted because it represents the velocity of

the flow with respect to the blade, rather than that of the blade with respect to the

flow. In the undeformed preconed blade coordinate system, the velocity VP can be

where V11 , V12 and V13 are the velocity components in the direction of êx , êy and êz ,

the unit vectors of the undeformed preconed blade coordinate system. The second

system, is therefore

where V11 , V12 and V13 are the components of the velocity VP , and λz is the induced

velocity in the z direction (the components of the induced velocity in the x and y

directions could be included, but are not used in the present study).

the air. The relative velocity of the flow with respect to the blade, VT F , is therefore

VT F = −VT (2.49)

52

which expressed in the same undeformed preconed coordinate system is

with

Vx = −V11 (2.51)

and so on.

For the aerodynamic calculations, however, the more traditional airflow veloc-

VA = UT eT + UP eP + UR eR (2.52)

considered positive for an airflow coming toward the leading edge of the airfoil, UR

is positive for an outboard flow and UP is defined as positive for a flow coming from

above.

The local velocity components can be used to calculate the local angle of

attack, αy , and yaw angle, γI , which are shown in Fig. 2.7, and which are

|UT |

cos γI = p 2 (2.53)

UT + UR2

−1 (UT tan θG − UP ) cos γI

αY = tan (2.54)

UT + UP tan θG cos2 γI

53

in which ψ is the blade azimuth angle, θT W is the built-in twist, ∆SP is the swash-

plate phasing angle, and φ is the elastic rotation of the blade section about the

can be approximated as the difference between the geometric and induced angles

−1 UP

αY ≈ θ − φ = θ − tan (2.56)

UT

With the angle of attack and the Mach number at any blade section, the

Two-dimensional quasi-steady aerodynamics (Ref. 94) are used for the blade

detail in Ref. 6, but a summary is included here for convenience. The basic expres-

sions for the distributed lift L and pitching moment M are (with the acceleration

terms ḧ and α̈ neglected for simplification purposes; therefore, the α̇ term is the

1

L = LQ + aρ(bR)2 V0 α̇ (2.57)

2

1

M = LQ xA − aρV0 α̇(bR − xA )(bR)2 (2.58)

2

where a is the lift-curve slope obtained from look-up tables as a function of the

angle of attack αY and the Mach number M , ρ is the air density, b is the non-

dimensional semi-chord length, R is the blade radius, α is the total pitch angle of

the blade section, which is θG − φ, where θG is the geometric pitch angle and φ is the

rotation of the cross section of the blade around the elastic axis, V0 is the oncoming

54

freestream flow defined as

q

V0 = UP2 + UT2 + UR2 (2.59)

xA is the blade cross-sectional aerodynamic center offset from the elastic axis (pos-

itive for aerodynamic center forward of the elastic axis) and α̇ is the time rate of

change of the total blade pitch angle, which corresponds to θ̇G − φ̇.

1 2 aα̇ c

LQ = ρV c CL + − xA (2.60)

2 0 V0 2

The lift equation (Eq. (2.57)), with the chord c substituted for the semi-chord

b, becomes

1

L = LQ + aρA V0 c2 α̇ (2.61)

8

1

D = ρA V02 cCD (2.62)

2

where CD is the steady drag coefficient obtained from look-up tables as a function

The aerodynamic lift L and drag D forces have to be transformed to the local

1 UT

fP = −L + DUP (2.63)

V0 cos γI

1

fT = [DUT + LUP cos γI ] (2.64)

V0

1 UP cos γI UR

fR = DUR + L (2.65)

V0 UT

55

along the eP , eT and eR unit vectors, respectively.

coned blade coordinate system by converting the force components in the blade

Z

pI = − ρ(aP + gkI )dA

A

Z

= − ρ (y0 ê0x + z0 ê0y ) × (aP + gkI ) dA

qI

A

56

where ρ is the mass density of the blade, y0 and z0 are the coordinates of the generic

mass point of the cross section A (see Fig. 2.5) and gkI is the contribution due to

gravity.

The inertia loads depend on the absolute acceleration of a point on the rotor

blade, aP , which is calculated by taking derivatives from the position vector of the

point, RP .

The position vector of a point on the blade, which is not necessarily on the

where the underlined terms represent the distance of the point from the elastic axis.

system by using the coordinate transformations presented in Eqs. (2.14) and (2.16),

as

RB = (e cos βp + u) + x0 +S21 y0 + S31 z0 êx +

v+S22 y0 + S32 z0 êy + w − sin βP +S23 y0 + S33 z0 êz (2.71)

The absolute velocity of the point is the same as that in Eq. (2.37), with the

exception of the RB vector which is defined in Eq. (2.71). The acceleration of the

d2 RCG ∂ 2 RB ∂RB

aP = + + 2ω × + ω̇ × (RB + RH ) +

dt2 ∂t2 ∂t

ω × [ω × (RB + RH )] (2.72)

57

where

∂ 2 RB ∂ 2 RB

∂RB

= + Ω̇ × RB + 2Ω × + Ω × (Ω × RB ) (2.73)

∂t2 ∂t2 R ∂t R

d2 RCG

= u̇F iB + v̇F jB + ẇF kB (2.74)

dt2

where u̇F , v̇F and ẇF are the linear acceleration components in the body fixed axes

system.

From the above expressions, one can obtain an expression for the absolute

d2 RCG

2

∂ RB

aP = + ω̇ × RH + ω × (ω × RH ) + + Ω̇ × RB +

dt2 ∂t2 R

∂RB

2Ω × + Ω × (Ω × RB ) + ω̇ × RB +

∂t R

∂RB

2ω × + Ω × RB + ω × (ω × RB ) (2.75)

∂t R

A more detailed expression of the acceleration terms can be found in Ref. 114.

The calculation of the structural loads on the main rotor blade is based on the

deflections in flap and lag bending, and torsion. A detailed derivation of the struc-

The resultant loads, expressed as operators on the first and second derivatives

58

with respect to the spanwise coordinate of the mode shapes, are

pII

Sy = Mz − S32 My (2.76)

pII

Sz = −(My − S23 Mz ) (2.77)

pISx = T (2.78)

I

qSx = Mx (2.81)

where the terms S12 , S13 , S23 and S32 are elements of transformation matrix

from undeformed to deformed coordinates given in Eq.2.16, and the terms S12,x ,

Z Z

F = T ê0x + Vy ê0y +

tdA Vz ê0z = (2.82)

A

Z Z

0 0 0

M = Mx êx + My êy + Mz êz = d × tdA (2.83)

A

where

with the stresses σxx , τxy and τxz being defined in detail in Ref. 114.

of a deformed rod (Ref. 115). The tension loads are derived in Refs. 6 and 114, and

59

in the blade preconed undeformed coordinate system they are

where T is the tension along the êx direction and S12 and S13 are elements of

Since the current study does not include axial degrees of freedom, axial dynamics

If a lag damper is included, the moment it generates about the hinge, MD , is given

by

The procedure to compute the aerodynamic, inertial, structural and tension loads for

the swept tip is similar to that used for the straight portion, but with the following

modifications:

• The loads on the swept tip are calculated in the local tip coordinate system.

For the calculation of the hub loads, they are transformed later to the blade

60

coordinate system before being added to the loads of the straight portion of the

blade. For the calculation of the blade natural frequencies and mode shapes,

these are not transformed, and the mass and stiffness matrices are calculated

• The elastic displacements of the points on the tip need to be in the local tip

coordinate system.

• The structural properties of a point in the swept tip need not be the same

as in the equivalent straight blade radial positions. The properties used for

the straight blade are approximations, since the actual blade of the UH-60 is

sheared. The original UH-60 data is used for the swept blade.

The position vector of a point on the tip section of the blade is given by:

where e1 is the blade root offset, xt is the point along the tip of the blade in local

axes, starting from the joint between the straight and swept portions of the blade,

nate system, which is the frame used for the calculation of the loads on the swept tip.

Using the transformations presented in Eqs.(2.14), (2.30) and (2.32), the position

61

vector becomes:

The velocity and acceleration are calculated by taking derivatives of the posi-

tion vector RP of Eq.(2.91) in the same way as for the straight blade, Eqs.(2.37) and

(2.75). The only difference is that the linear and angular velocities and accelerations

are transformed to the tip preconed undeformed coordinate system using Eq.(2.29).

θG = (θ0 + θ1c cos(ψ + ∆SP ) + θ1s sin(ψ + ∆SP ) + θBJ ) cos Λ + θBT (xt ) (2.92)

where θBJ is the value of the built-in twist at the junction of the straight and swept

portions of the blade, θBT is the pretwist of the swept tip with respect to the junction

to the straight portion of the blade, ∆SP is the swashplate phasing angle, and xt is

The equations representing the motion of the rotor blades are a system of cou-

pled non-linear partial differential equations with time varying coefficients. These

equations using a finite element discretization (see Section 4.2). Moreover, a modal

62

coordinate transformation is performed in order to reduce the number of degrees of

freedom as well as the number of equations for the dynamics of each rotor blade (see

Section 4.4). The result is a system of nonlinear, coupled, second order ordinary

symbolic form

q̈ = fq (q̇, q) (2.93)

63

2.5 Fuselage equations of motion

The main assumption in the formulation of the fuselage equations of motion is that

The non-linear force and moment equilibrium equations, which are expressed

R

X = mu̇F + m(qwF − rvF ) + mg sin θF (2.94)

m0

R

Y = mv̇F + m(ruF − pwF ) − mg cos θF sin φF (2.95)

m0

R

Z = mẇF + m(pvF − quF ) − mg cos θF cos φF (2.96)

m0

R3

L = Ixx ṗ − Ixy q̇ − Ixz ṙ − Iyz (q 2 − r2 ) − Ixz pq + Ixy pr

m0

−(Iyy − Izz )qr (2.97)

R3

M = Iyy q̇ − Ixy ṗ − Iyz ṙ − Ixz (r2 − p2 ) − Ixy qr + Iyz pq

m0

−(Izz − Ixx )pr (2.98)

R3

N = Izz ṙ − Ixz ṗ − Iyz q̇ − Ixy (p2 − q 2 ) − Iyz pr + Ixz qr

m0

−(Ixx − Iyy )pq (2.99)

In Eqs. (2.94)-(2.99), the mass and inertia terms have been non-dimensionalized

with respect to the main rotor radius R and a reference blade mass m0 .

The force and moment components on the left hand side of Eqs. (2.94)-(2.99)

are the externally applied loads at the center of gravity of the body, described below.

The right hand side represents inertia forces and moments due to the rigid body

The total applied forces and moments are the sum of contributions from the

64

main rotor, tail rotor, fuselage and empennage, that is:

X = XM R + XT R + XF + XV + X H (2.100)

Y = YM R + YT R + YF + YV + YH (2.101)

Z = ZM R + ZT R + ZF + ZV + ZH (2.102)

L = LM R + LT R + LF + LV + LH (2.103)

M = MM R + MT R + MF + MV + MH (2.104)

N = NM R + NT R + NF + NV + NH (2.105)

where the subscript M R denotes the main rotor, T R the tail rotor, F the fuselage,

The fuselage rigid body equations are completed with three equations that

relate the aircraft angular rates p, q, r, and the rates of change of the Euler angles

cos ψF sin φF

ψ̇F = r +q (2.108)

cos θF cos θF

The following subsections describe the components of the external forces and

moments applied to the center of gravity of the body that appear in Eqs.(2.100)

through (2.105).

65

2.5.1 Main rotor loads

The contributions from the main rotor to the fuselage loads consists of the sum

of the distributed aerodynamic and inertial loads, integrated along the span of the

blade to obtain the loads at the hinge. The integrated loads are formulated in the

undeformed preconed coordinate system (Section 2.2.2). These forces and moments

are given by

Z 1

FR = (pA + pI )dx0 (2.109)

e

Z 1 Z 1

MR = (qA + qI )dx0 + RC × (pA + pI )dx0 + MD (2.110)

e e

and (2.87), respectively, and e is the hinge offset. RC is the position vector of a point

on the deflected elastic axis from the hub in the undeformed preconed coordinate

The main rotor loads are transformed to equivalent loads at the center of

For an articulated rotor configuration, the flap and lag aerodynamic or inertial

blade moments are not transferred through the hinge. The forces generated by the

rotor are obtained by integrating the aerodynamic and inertial distributed loads

over the elastic portion of the blade. For a single rotor blade, these forces are

R

1

X (p + pIx )dx0

MR Re Ax

−1 −1 −1 1

FM R = YM R = [TSB ] [TRS ] [TP R ] e

(pAy + pIy )dx0 (2.112)

Z

1 (p + p )dx

R

MR e Az Iz 0

66

where the coordinate transformation matrices [TSB ], [TRS ] and [TP R ] are defined,

The moment vector at the center of gravity of the body, for a single blade, is

L MDx

MR

−1 −1 −1

MM R = MM R = [TSB ] [TRS ] [TP R ] MDy

R 1

NM R

0

(q Az + q Iz )dx 0 + MDz

+ R H × FM R

The first term indicates that the lag damper moments are all transferred

through the hinge, but only the pitching moments of the distributed inertial and

aerodynamic loads are. The second term transforms the forces at the blade hinge

into moments at the hub, where RC is the position vector from the hub to the hinge

RC = eêx (2.114)

The third term transforms the forces at the hub into moments at the center of

gravity of the body; the moment arm, RH , is the distance of the hub from the

If the swept tip model is included, the loads calculated in the tip local coordinate

global blade coordinate system before they can be integrated with the loads on the

67

The local aerodynamic forces and moments for a point in the swept tip, pLAx ,

L L

, qAy L

and qAz , are calculated as described in Sections 2.4.2

and 2.4.7.

pG L L

Ax = pAx cos Λ + pAy sin Λ (2.115)

pG L L

Ay = pAy cos Λ − pAx sin Λ (2.116)

pG L

Az = pAz (2.117)

and

G L L

qAx = qAx cos Λ + qAy sin Λ (2.118)

G L L

qAy = qAy cos Λ − qAx sin Λ (2.119)

G L

qAz = qAz (2.120)

Similarly, for the inertial forces and moments, the transformation becomes:

pG L L

Ix = pIx cos Λ + pIy sin Λ (2.121)

pG L L

Iy = pIy cos Λ − pIx sin Λ (2.122)

pG L

Iz = pIz (2.123)

and

G L L

qIx = qIx cos Λ + qIy sin Λ (2.124)

G L L

qIy = qIy cos Λ − qIx sin Λ (2.125)

G L

qIz = qIz (2.126)

68

2.5.2 Fuselage aerodynamic loads

The aerodynamic forces and moments acting on the body of the fuselage are ex-

loads are based on the freestream velocity at the fuselage aerodynamic reference

point (Ref. 32), with a correction factor to take into account the interference of the

main rotor

uF = uB + yf rB − zf qB + uinf (2.127)

vF = vB + zf pB − xf rB + vinf (2.128)

wF = wB + xf qB − yf pB + winf (2.129)

where xf , yf and zf represent the components of the position vector from the center

of gravity of the body to the aerodynamic reference point of the fuselage in the

body axis system. The components uinf , uinf and uinf are the interference velocities

based on the main rotor downwash, tip speed and wake skew angle, and are based

vinf = 0 (2.131)

with β1c being the longitudinal tilt of the tip path plane, given in Eq. (2.20); v0 is

the main rotor average downwash, which is taken as the constant inflow coefficient

λ0 when dynamic inflow is used to calculate the induced velocities; and χ is the

69

rotor wake skew angle, which is

uS

χ = tan−1 + β1c (2.133)

|v0 − wS |

where uS and wS are the freestream velocity components taken in the shaft fixed

The angles of attack and sideslip of the fuselage are defined as:

wF

αF = tan−1 (2.134)

|uF |

vF

βF = tan−1 p 2 (2.135)

vF + wF2

1 R2

q̄F = ρ (u2F + vF2 + wF2 ) (2.136)

2 m0

For the UH-60A, the non-linear fuselage aerodynamic coefficients are defined

CY f = CY f (|βF |) (2.138)

βF

CRf = − CRf (|βF |) (2.140)

|βF |

βF

CM f = CMα f (αF ) − CM f (|βF |) (2.141)

|βF |

CN f = CN f (−βF ) (2.142)

70

The non-dimensional fuselage aerodynamic loads in the wind axes system are

These loads need to be transformed from the wind axes system to the body

axes system, using the angles of attack and sideslip at the aerodynamic reference

point of the fuselage. The transformation is similar to the transformation from wind

cos αF cos βF cos αF sin βF − sin αF

TFB = (2.145)

sin βF cos βF

sin αF cos βF − sin αF sin βF cos αF

so that

i i

B F

−1

jB = [TFB ] jF (2.146)

k

k

B F

Once resolved at the center of gravity of the body, the forces and moments

X −q̄F CDf

F

FF = YF = [TF B ]−1 FwF = [TF B ]−1 −q̄F CY f (2.147)

Z −q̄ C

F F Lf

and

L

F

MF = MF = [TF B ]−1 MwF + xF × FF (2.148)

N

F

where xF is the position vector of the fuselage aerodynamic reference point from

xF = xf iB + yf jB + zf kB (2.149)

71

In the body axes system, the non-dimensional aerodynamic forces at the center

of gravity are:

X q̄ C

F F Xf

FF = YF = q̄F CY f (2.150)

Z

q̄ C

F F Zf

L q̄ C

F F Lf

MF = MF = q̄F CM f + xF × FF (2.151)

N q̄ C

F F Nf

The aerodynamic loads acting on the horizontal and vertical tail surfaces contribute

to the total loads on the helicopter. The loads are calculated with velocities at the

uH = KH uB + xH × ω + uinH (2.152)

uV = KV uB + xV × ω + uinV (2.153)

KH and KV , which are obtained empirically, determine the dynamic pressure losses

at the horizontal and vertical tail (Ref. 32); the vector products xH × ω and xV × ω

are the velocity components due to the rotation of the body, and uinH and uinV are

the interference velocity components due to the downwash of the main rotor and

1 R2

q̄H = ρ (u2H + vH2 2

+ wH ) (2.154)

2 m0

1 R2

q̄V = ρ (u2V + vV2 + wV2 ) (2.155)

2 m0

72

The interference velocities are a function of the main rotor downwash, tip

where v0 is the main rotor downwash (which, like in the case of the fuselage loads,

is equal to the constant inflow coefficient when dynamic inflow is used, but zero

when the free wake model calculates the induced velocities), β1c is the longitudinal

tilt of the tip path plane, given in Eq. (2.20), and χ is the rotor wake skew angle

Eq.(2.133). The functions νxw H (β1c , χ), νzw H (β1c , χ), νxw V (β1c , χ) and νzw V (β1c , χ)

The angles of attack and sideslip at the horizontal and vertical tail are:

wH

αH = tan−1 + θ0H (2.158)

|uH |

v

βH = tan−1 p H (2.159)

2 2

vH + wH

wV

αV = tan−1 (2.160)

|uV |

v

βV = tan−1 p V (2.161)

vV + wV2

2

The angles of attack are positive nose up, the angle of sideslip are defined as positive

nose right and θ0H is the variable pitch angle of the horizontal tail, which depends

on the flight speed and is adjusted by the flight control system (Ref. 32).

The empennage aerodynamic loads acting of the UH-60A are defined in the

local wind axes system, the lift and drag aerodynamic coefficients at the tail surfaces

73

being:

The total aerodynamic loads in the local wind axes systems of the empennage

are:

where SH and SV are the surface areas of the horizontal and vertical tail respectively.

A transformation is necessary to convert these forces, from the local wind axes

system of the horizontal (iH , jH , kH ) and vertical (iV , jV , kV ) tails, into the body

cos(αH − θ0H ) cos βH cos(αH − θ0H ) sin βH − sin(αH − θ0H )

THB = − cos βH (2.168)

sin βH 0

sin(αH − θ0H ) cos βH sin(αH − θ0H ) sin βH cos(αH − θ0H )

so that:

i i

B H

−1

jB = [THB ] jH (2.169)

k

k

B H

74

the body axis at the center of gravity of the body are

XH

FH = YH = [THB ]−1 FwH (2.170)

Z

H

and

L

H

MH = MH = xH × FH (2.171)

N

H

with xH being the position vector of the horizontal tail aerodynamic reference point

xH = xH iB + yH jB + zH kB (2.172)

cos αV cos βV cos αV sin βV − sin αV

TVB = − cos βV (2.173)

sin βV 0

sin αV cos βV sin αV sin βV cos αV

so that

i i

B V

jB = [TVB ]−1 jV (2.174)

k

k

B V

And the forces and moments at the center of gravity of the body are:

X

V

FV = YV = [TVB ]−1 FwV (2.175)

Z

V

and:

L

V

MV = MV = x V × FV (2.176)

N

V

75

where xV is the position vector from the center of gravity of the body to the vertical

xV = xV iB + yV jB + zV kB (2.177)

The tail rotor model is based on a modification of the simplified closed-form Bailey

solution (Ref. 116). At the tail rotor, the velocity of the airflow in the body fixed

axes system is

uT R = uB + xT R × ω + uinT R (2.178)

where uinT R is the interference velocity caused by the fuselage and main rotor wake

and is a function of the main rotor downwash, tip speed and wake skew angle:

where the functions νxw T R (β1c , χ) and νzw T R (β1c , χ) are read from look-up tables (Ref. 32).

The position vector from the hub of the tail rotor to the center of gravity of

xT R = xT R iB + yT R jB + zT R kB (2.180)

The velocity components at the tail rotor hub in the body fixed coordinate

system are

uT R = uB + yT R rB − zT R qB + uinT R (2.181)

vT R = vB + zT R pB − xT R rB + vinT R (2.182)

wT R = wB + xT R qB − yT R pB + winT R (2.183)

76

These velocities need to be expressed in the local tail rotor coordinate system.

For this transformation, two rotations are necessary, one about the xB axis by the

tail rotor cant angle, ΓT R , and the second about the new z axis by the tail rotor

cos ΛT R − sin ΓT R sin ΛT R cos ΓT R sin ΛT R

TTB = (2.184)

0 cos ΓT R sin ΓT R

− sin ΛT R − sin ΓT R cos ΛT R cos ΓT R cos ΛT R

The velocity components in the local tail rotor coordinate system become:

u u

tl TR

vtl = [TTB ] vT R (2.185)

w

w

tl TR

R2

2 2

Ttl = 2ρ π R̄t νt vT t Ω̄t R̄t Kblk (2.186)

m0

where R̄t and Ω̄t are the non-dimensional tail rotor radius and rotational speed re-

spectively, non-dimensionalized with respect to the main rotor radius and rotational

speed each; νt is the tail rotor induced velocity, Kblk is an empirical tail rotor block-

age factor to account for the presence of the vertical tail (Ref. 32) and vT t is the

total speed of the airflow at the tail rotor hub, which is given by:

p

vT tl = µ2t + λ2t (2.187)

where

q

µtl = u2tl + vtl2 (2.188)

77

λt = wtl − νt (2.189)

2

1 2 3 R

Qtl = ρ Ω̄t R̄t π R̄t (2.190)

2 m0

The thrust and torque loads in the local tail rotor coordinate system are:

After transforming these loads at the center of gravity of the body obtained

using the transformation from tail rotor coordinate system to body fixed system,

the resultant forces and moments at the center of gravity due to the tail rotor are:

X

TR

FT R = YT R = [TTB ]−1 Ftl (2.193)

Z

TR

and

L

TR

MT R = MT R = [TTB ]−1 Mtl + xT R × FT R (2.194)

N

TR

The tail rotor inflow dynamics are calculated with a 1-state Pitt–Peters dynamic

inflow model (Ref. 24). Therefore, only one constant inflow state is used, the sine

and cosine terms are zero, and only one equation is needed to represent the inflow

78

1

τt ν̇t + νt = Lt CTt (2.195)

Ωt

1 4

τt = (2.196)

vTt 3π

1

Lt = (2.197)

2vTt

vTt is the total induced velocity at the tail rotor, given by:

q

vTt = (u2t + λ2t ) (2.198)

Ttl m0 Ω2 R2

CTt = (2.199)

ρπΩ2t Rt4

where Ωt is the tail rotor rotational speed, Rt is the tail rotor radius and Ttl is the

The previous sections have described the different components necessary to build

the complete equations of motion of the helicopter. These equations are formulated

ẏ = g(ẏ, y, u ; t) (2.200)

where y and ẏ are the state vector, Eq.(2.4), and its derivative, respectively, u,

79

1. The nonlinear, rigid body equations of motion for the helicopter, known as

the Euler equations. These are a set of 9 equations representing the force and

relationship between the angular rates and the derivatives of the Euler angles,

4. If dynamic inflow is used to calculate the main rotor induced velocities, the

equations of the Peters–He dynamic inflow model are provided (Ref. 36). If in-

stead the inflow is calculated with the free wake model, the free wake equations

80

XB

CG

XI

YB

YI ZB

ZI

81

XB

φ

θ XI

ψ

ψ

θ

φ φ

θ

ZB

YI

ψ

YB ZI

Figure 2.2: Euler angles and rotations from the inertial to fuselage coordinate sys-

82

XB

YB

α w = Vsinα

β

u = Vcosαcosβ

v = Vcosαsinβ

ZB

Figure 2.3: Relationship between velocity vector defining the wind coordinate sys-

83

k

~ YS

~j

i

~

ψ

XS

ZS

84

`

^

eZ

` `

yO ^ `

eY ^

eZ ^

` eX

zO

^

eY

^

eX

Def ormed

Blade

k

~ ^

eZ i

~

^

~j eY

Undef ormed

βp Blade

85

jT

jH ^

eX

Def ormed

Blade iT

Λ

iH

k

~ i

^

eZ ~

Undef ormed

^ Blade

eY

~j βp

86

FT VO

FP UP

tip FR root

φY

UT γ

I

UR

α Y = θY + φY

Plane of Positive

Blade Pitch, θ

θ

tip root

Blade Span

Axes θY

UT

γ

I Plane of Zero

UR

Blade Pitch

Figure 2.7: Definition of the sectional aerodynamic angles, the yaw angle γI and the

87

Chapter 3

3.1 Overview

The wake of a helicopter rotor is comprised mostly of the strong vortices that are

released from the tips of its blades. These vortices form quickly behind the rotor

blades and are convected below the rotor in a helical structure in hover, and are

skewed back with increasing forward speeds (Ref. 3). Numerical vortex models

describe the rotor wake by tracking the tip vortices through the flow field (Ref. 3).

While there are several approaches to modeling the tip vortices, the free wake models

used in the present study discretize the vortex filaments with straight segments,

whose strength and location constitutes part of the problem being solved. By virtue

of the vorticity transport theorem, each vortex filament can be modeled and its

location determined. With the position and strength of the vortex filaments known,

one can compute the induced velocity field at the rotor by means of the Biot–Savart

law. Free wake models are a particular type of vortex model in which the vortex

filaments are allowed to distort freely under their self and mutual interaction, as

88

rigid or prescribed wakes. The exact position of each filament is calculated as part

There are two main categories of free wake methodologies: relaxation (or iter-

ity, which makes them more appropriate for steady-state solutions. Time-marching

schemes are potentially subject to numerical instabilities, but are not restricted by

The free wake model used in this study is the time-accurate maneuver free wake

by Ananthan and Leishman (Refs. 5, 117). The distortions of the wake geometry due

to maneuvers are taken into account without a priori assumptions on the geometry.

The free wake model used in the present study is comprised of two parts:

the near wake, which is assumed planar and with a fixed angular length, and the

tip vortex that constitutes the free wake and which extends beyond the near wake

3.2 Assumptions

The following assumptions are associated with the free wake model and the coupling

the free wake model to the rest of the flight dynamic model.

89

compressible flow (i.e., a potential flow).

single vortex released from the blade elastic axis at the blade tip. This as-

sumption may be questionable at high speeds or through the vortex ring state,

but those are aerodynamic issues, and for flight dynamics considerations, the

3. The bound circulation and lift are assumed constant over each blade segment

4. Thin airfoil theory is assumed in the relation between lift and circulation in

5. The initial tip vortex strength is taken to be the maximum bound circulation

Additionally, the following assumptions are taken in the coupling of the free

1. Within the free wake model, it is assumed that the blade is rigid and straight

with a single flap hinge at the axis of rotation. An “equivalent” flap angle is

2. The free wake model provides all three spatial components of the induced

velocity in the free wake blade preconed coordinate system (see Section 3.3.2),

90

however the x and y components are set to zero and only the z component (i.e.,

the component along the zP axis) is used for the calculation of the aerodynamic

loads. This assumption can be safely done for most flight conditions, for which

However, for the vortex ring state conditions this may not be true and this

3. No lag and torsion motion is included in the free wake model. However, the

effect of flap, lag and torsion is included in the blade sectional velocities that

are provided by the flight dynamics model to the free wake model.

4. The effect of the downwash of the rotor on the fuselage and empennage is not

included.

The free wake model is formulated in a wind coordinate system (Refs. 12, 15, 40),

In the global wake coordinate system, in which the free wake model is formulated,

the x-axis is aligned with the freestream flow and is positive aft, the y-axis points to

starboard and the z-axis points up. For a wind tunnel trim problem, the x-axis is

91

aligned with the longitudinal axis of the wind tunnel. In hover, where the freestream

velocity is zero, the x-axis is assumed to point along the longitudinal axis of the

helicopter.

Figure 3.1 illustrates this coordinate system. The longitudinal shaft tilt angle,

αS , is the angle that the z-axis makes with the rotor shaft, and also indicates the

angle of attack of the rotor hub with respect to the freestream flow. αS is defined

as positive with the shaft tilted aft so that the thrust vector is also tilted in the

aft direction. The axes of the global coordinate system are xGW , yGW , zGW , with

The free wake blade preconed coordinate system is the system in which the free wake

calculates the induced velocities. It has its origin at the hub and its unit vectors are

iP , jP , kP with the x-axis pointing outboard along the blade, the y-axis pointing in

The transformation from the global-fixed frame to the free wake blade preconed

rotating frame requires first a rotation of the longitudinal shaft tilt αS about the

global fixed yG W -axis, followed by a rotation of the azimuth angle ψ about the new

z-axis (aligned with the rotor shaft), and a rotation of the precone angle βP about

the new y-axis (normal to the rotor shaft and pointing in the blade lead direction).

92

cos αS cos ψ cos βp − sin αS cos ψ cos βp

sin ψ cos βp

+ sin αS sin βp + cos αS sin βp

[TPG ] = − cos αS sin ψ (3.1)

cos ψ sin αS sin ψ

− cos αS cos ψ sin βp

sin αS cos ψ sin βp

− sin ψ sin βp

+ sin αS cos βp + cos αS cos βp

so that:

i i

P GW

jP = [TPG ] jGW (3.2)

k

k

P GW

3.3.3 Coupling of the flight dynamic and the free wake co-

ordinate systems

The flight dynamics model and the free wake model are each formulated in a slightly

different frame of reference. At the interface between both models, the necessary

transformations must take place, in both directions, in order to ensure the data

There are four sets of information that are exchanged between the two models,

Body velocities and rates The freestream velocity and the body angular rates,

which are in the body axis system iB , jB , kB , in the flight dynamics model

(Eq.(2.7)), need to be converted to the global wake axis system iGW , jGW ,

Flap angle The flap angle at the tip of the blade needs to be provided by the

93

flight dynamics model to the free wake model. Both the rotor-fuselage model

and the free wake model consider the flapping angle β positive for an upward

Blade velocities The velocity of the flow seen by the blade at the different sections

along the blade span needs to be provided by the flight dynamics model to

the free wake model. This velocity includes the freestream velocities at the

blade section, the velocities due the angular rotation of the helicopter and

the blade and the velocities due to the elastic deformation of the blade (see

Section 4.5.5 for further details). In the flight dynamics model, these velocities

are calculated in the blade undeformed preconed coordinate system, êx , êy ,

êz (Eq.(2.15)). In the free wake model, these velocities are required in the

Induced velocities The inflow distribution calculated in the free wake model needs

to be provided to the rest of the flight dynamic model, and therefore needs

êx , êy , êz (Eq.(2.15)). These two coordinate systems are coincident, so no

transformation is required.

94

3.3.4 Body-fixed to global wake coordinate systems

The transformation from the body-fixed to the wake global wake reference frames

−1 0 0

TGB = 0 1 0 (3.3)

0 0 −1

so that:

i i

GW B

jGW = [TGB ] jB (3.4)

k

k

GW B

Therefore, the velocity components and the angular rates of the body can be

converted from the body-fixed system to the global wake axis system as follows:

u −uB

GW

vGW = vB (3.5)

w

−w

GW B

and:

p −pB

GW

qGW = qB (3.6)

r −r

GW B

The behavior of the vortex filaments is described by the equation of vorticity trans-

port (Ref. 7)

dr (ψ, ζ)

= V (r (ψ, ζ)) (3.7)

dt

∂r (ψ, ζ) ∂r (ψ, ζ) 1

+ = V (r (ψ, ζ)) (3.8)

∂ψ ∂ζ Ω

95

where r(ψ, ζ) defines the position of a point on the vortex filament and V(r(ψ, ζ))

The derivatives on the left hand side of Eq. (3.8) are calculated using finite

difference approximation (Ref. 7). For this purpose, the wake is discretized both in

space and time. A schematic of the wake discretization is shown in Fig. 3.3. In the

time domain, ψ, the rotor azimuth is divided into Nψ angular steps, each of size ∆ψ.

Since there is one filament released at each time step, the total number of vortex

filaments representing the tip vortex geometries at the entire set of azimuth angles

is

2π

Nψ = (3.9)

∆ψ

resolution is ∆ζ. The total number of straight line vortex segments in each filament

is

ζmax

Nζ = (3.10)

∆ζ

and the total number of collocation points for each vortex filament is Nζ + 1.

In Fig. 3.3, the index j in the collocation points is used as a reference to the

azimuth angle ψ and the index k references the location of the collocation point in

The right hand side of Eq. (3.8) is determined by the total velocities at the

collocation point, which comprise the free-stream velocity, the velocities induced by

all the other vortex filaments and the blades, plus the velocities due to maneuvering.

96

3.4.1 The time marching scheme

The time-marching free wake model is based on the solution of the governing partial

differential equations at the midpoints of the grid cell defined by the discretizations

Dζ ≈ =

∂ζ

r(ψ + ∆ψ, ζ + ∆ζ) − r(ψ, ζ + ∆ζ) + r(ψ + ∆ψ, ζ) − r(ψ, ζ)

(3.11)

2∆ζ

∂r(ψ + ∆ψ/2, ζ)

Dψ ≈ =

∂ψ

3r(ψ + ∆ψ, ζ) − r(ψ, ζ) − 3r(ψ − ∆ψ, ζ) − r(ψ − 2∆ψ, ζ)

(3.12)

4∆ψ

The velocity field at each segment of the vortex filament is composed of the free-

stream velocity, the velocities self and mutually induced by all the vortex filaments

and the blades, plus any additional velocities due to maneuvering (Ref. 7).

97

The velocity term in the RHS of Eq. (3.8) is therefore given as

The velocities due to the maneuvering of the helicopter are calculated with

the cross product of the angular rates by the position vector of the given point from

where the subscript G denotes the global wake coordinate system described in Sec-

tion 3.3.1.

The most difficult and expensive term to compute is the one corresponding to

the induced velocities. The Biot–Savart law is used to calculate the velocity induced

at a point located at position r relative to the vortex element dl, for each filament

dlj × ri

Z

Γj

Vind = (3.16)

4π |ri − rj |3

Two aspects of the vortex model that are important in the calculation of the

induced velocities are the tangential velocity profile and the diffusion of the vortex

a “solid-body” rotation and an outer region that simulates a potential vortex profile.

98

Ref. 15 shows that the tangential velocity vθ of the rotor tip vortices can be closely

approximated by

Γr

vθ (r) = p (3.17)

2π rc4 + r4

where Γ is the vortex circulation strength, r is the distance at which the tangential

radius rc in which the core radius grows as function of the vortex age (Ref. 40)

r

4αδνζ

rc (ζ) = rc20 + (3.18)

Ω

where α is the Oseen parameter and has a value of 1.25643, ν is the kinematic

viscosity of air, ζ is the vortex age in radians, Ω is the rotational speed of the rotor,

and δ is an “eddy” or turbulent velocity coefficient that determines the rate at which

the vortex core grows with time, which is determined empirically (Ref. 15).

Including the tangential velocity profile model, Eq. (3.17), and vortex diffusion,

Eq. (3.18), into the Biot–Savart law, the induced velocity due to the vortex element

is

dlj × r

Z

Γj h

Vind = p (3.19)

2π rc4 + h4 |ri − rj |3

where h is the perpendicular distance of the evaluation point from the influencing

vortex element. The total induced velocity at a point in the flow field results of

The present model uses realistic values for the vortex core growth obtained

from experiments (Ref. 119) rather than an arbitrary core value to improve conver-

99

3.6 Bound circulation and the near wake problem

The bound circulation is obtained using a Weissinger-L lifting surface model (Ref. 51),

which is a type of lifting surface model that relates the blade lift to the bound cir-

culation through the Kutta–Joukowski theorem. The bound circulation along the

blade determines the strength of the trailed wake vortices, which in turn influence

In this model, the blade is discretized into NS spanwise segments and only one

chordwise segment, as shown in Fig. 3.4 (Refs. 15, 40). At each segment, a control

point is located at 3/4 of the chord, while the bound circulation is located at the

quarter-chord location and is assumed constant along the segment. The difference

endpoints, with a vortex strength equal to the difference between the two segments

bound vortex strengths. These trailed vortices comprise the near wake, which is

assumed planar and with a fixed angular length. The tip vortex that constitutes

the free wake extends beyond the near wake with a strength equal to the maximum

To solve for the strength of the bound circulation, one must satisfy the flow

tangency condition (Ref. 120), which means that the flow cannot go through the

rotor blade, at least at the 3/4-chord point. Therefore, the component of the incident

velocity normal to the spanwise segment must be zero at that point (Ref. 40):

Vi · ni = 0 (3.20)

The total velocity at any point in the blade, Vi , is composed of the free-

100

stream velocities at the given point, the velocities due to the blade motion and

flexibility, the local velocities due to the maneuvering of the helicopter and the

induced velocities, which have three sources, the bound circulation, the near wake

The bound and near wake velocities can be expressed in terms of the strength

NS

X

VBi = Ibi,j Γj

j=1

NS

X

VN W i = IN W i,j Γj (3.22)

j=1

with i going from 1 to NS . The terms Ibi,j and IN W i,j are the bound and near wake

pressing in the form of a linear system of equations, yields the governing equation

NS

X

Ibi,j + IN W i,j Γj = {(V∞ + VBM + Vman + VF W ) · n}i (3.23)

j=1

The velocity induced by the trailed vortices, VF W , is calculated by the free wake

model. Since this velocity depends of the strength of the bound vortex itself, the the

circulation is updated with the new inflow at each iteration of the solving process

until both converge to a solution. The stream velocities at the control point, V∞ , are

provided by the flight dynamics model and include the velocity due to the translation

101

of the helicopter. The velocity due to the blade motion and flexibility VBM and

the velocities due to the rotation of the helicopter Vman , to account for the effect of

In order to account for the tip sweep in the blade, several modifications need to be

The blade control points on the free wake model, which are located at the

center of each blade segment, need to be moved backwards in the lag direction for

those points located on the swept tip. The distance they are displaced is an amount

equal to the product of the distance from the joint between the straight and swept

parts of the blade, xiJT , and the tangent of the sweep angle, ∆, i.e.,

The near wake, which extends for a fixed angular length behind the blade,

needs to be extended further in the region of the swept tip. For this purpose, the

location of the end points of the near wake at each radial station in the tip is

extended with the same approach as that used for the blade control points.

The release point of the vortex filament occurs at the end of the swept portion

of the blade.

The velocities seen by each blade section, used for the computation of the

circulation with the Weissinger-L model, are provided in a reference frame perpen-

102

The induced velocities are provided back to the rotor-fuselage model at the

103

T

ZG

V

βo

XG

βo

αs

104

Tip Flap

Displacement

Equivalent

St raight

Blade

Def ormed

Equivalent Flap Angle Blade

Based on Tip

Displacement

βo

Tip Displacement Tot al Flap

Hub Plane Due t o Precone Displacement

Rot or Hub Undef ormed From Hub Plane

Blade

Figure 3.2: Definition of equivalent tip flapping angles for the free wake model.

105

z y

q

Blade, N

ψ

approximation ζ

Ω

r l +1

p Blade, N-1

Γv x

Lagrangian l +2

markers

h

Γv Curved vortex

filament

Induced velocity from

element of vortex trailed

by blade N-1

106

Rotational Bound circulation

axis Γj-1 Γj Γj+1

z (loading)

distribution

y

Ω

x

at 1/4-chord

Control point

Γj-1 − Γj at 3/4-chord

Γj − Γj+1

Trailed circulation

in near-wake L-shaped vortex elements

107

Chapter 4

Solution methods

This chapter describes the different solution methodologies and techniques used in

the present study. The chapter starts with the procedure to solve the equations

of motion. Next is the description of the finite element method used to model the

blade, both in its straight and swept configurations. Following is the approach for

the calculation of the blade mode shapes and natural frequencies of these two blade

The last two sections describe, respectively, the trim methodology to obtain

the steady state solution of the helicopter, with the inclusion of the time marching

free wake model, and the method for the integration of the equations of motion.

In Eq.(2.200), the derivative terms, which are primarily associated with the main

rotor inertia, appear both on the right- and the left-hand-side of Eq.(2.200). This

requires that the derivative terms on the right-hand-side be identified and moved to

108

the form

f (ẏ, y, u ; t) = 0 (4.1)

With suitable manipulations and appropriate ODE solvers, the two approaches are

essentially equivalent. Both have been used to obtain the results of this dissertation.

The first approach is based on rewriting the equations of motion in the form

can be seen that, with the exception of some very small terms, they are linear in

gI (ẏ ; t) = EC ẏ (4.3)

finite difference approximations and gN are the equations of motion without the

ẏ = EC ẏ + gN (y, u ; t) (4.4)

or

ẏ = f1 (y, u ; t) (4.6)

109

Alternatively, the model can be simply expressed in implicit form (Ref. 111):

f (ẏ, y, u ; t) = 0 (4.7)

Because Eq.(4.7) does not require any preliminary manipulations, this is the

recommended approach.

The blade equations of motion, are a set of partial differential equations and need to

flight dynamic model. This is obtained through a finite element approach, which

removes the spanwise coordinate. The following finite element analysis is based on

For each finite element in the analysis, there are 11 degrees of freedom: the

flap and lag bending displacement and their slope at each end of the element, and

the torsional rotations at the two ends of the element and at the element mid-point.

Figure 4.1 depicts the nodal degrees of freedom in each element, in which φ, v and

The lag, flap, and torsion degrees of freedom for each element are, respectively

yφ = { φ0 φ1 φ2 }T (4.10)

110

where v0 , vx,0 , w0 , wx,0 are the lag and flap displacement and rotation at the

inboard end, v1 , vx,1 , w1 and wx,1 are the lag and flap displacement and rotation at

the outboard end, φ0 is the torsional rotation at the inboard end, φ1 is that at the

center and φ2 at the outboard end of the element. The complete vector of degrees

yi = { y v yw yφ } T (4.11)

Figure 4.2 shows how the local degrees of freedom of each finite element are

assembled to build the entire blade, for the case of the straight blade configuration.

The global degrees of freedom for the blade with the swept tip element are shown in

Fig. 4.3. The total number of degrees of freedom is 5 + 6Ne where Ne is the number

The degrees of freedom for the entire blade are assembled into the vector, yn :

yn = [ vn wn φn ]T (4.12)

where

φn = [ φ0 φ1 . . . φ2Ne ]T (4.15)

The flap, lag and torsion displacements at any point within each element

are reconstructed from the nodal degrees of freedom using Hermite interpolation

111

polynomials. For flap and lag, the polynomials are

T

1 − 3η 2 + 2η 3

η(1 − 2η + η 2 )l

Hv (xe ) = Hw (xe ) = (4.16)

3η 2 − 2η 3

η(−η + η 2 )l

1 − 3η + 2η 2

Hφ (xe ) = 4η − 4η 2 (4.17)

−η + 2η 2

where l is the length of the element, and η = xe /l, where xe is the distance from

the inboard node of the element. Therefore η = 0 represents the inboard node and

Using the Hermite polynomials, the flap, lag and torsion deflections at any

112

With the displacement quantities and their derivatives known at any point in

the blade, the distributed aerodynamic, inertial, tensile and structural loads can be

p (x , t) H (x ) p (t)

Iy e v e

Iiv

Z li

pIi (t) = pIz (xe , t) Hw (xe ) dxi = pIiw (t) (4.22)

0

q (x , t) H (x )

p (t)

Ix e φ e Iiφ

where pIy , pIz and qIx are components of the distributed inertial loads (Eqs.(2.68)

and (2.69) in Section 2.4.3). The terms pIiv , pIiw and pIiφ are the nodal load vectors

Similarly, the aerodynamic nodal loads, pA for the ith element can be calcu-

lated:

pAy (xe , t) Hv (xe ) p (t)

Z li Aiv

pAi (t) = pAz (xe , t) Hw (xe ) dxi = pAiw (t) (4.23)

0

q (x , t) H (x )

p (t)

Ax e φ e Aiφ

where pAy , pAz and qAx are the components of the blade section aerodynamic loads

The structural nodal load vector for the i-th element is given as

pISy (xe , t) Hv,x (xe ) + pII

Sy (xe , t) Hv,xx (xe )

p (t)

Siv

Z li

pSi (t) = pISz (xe , t) Hw,x (xe ) + pII

Sz (xe , t) Hw,xx (xe )

dxi = pSiw (t)

0

q (x , t) H (x ) + q I (x , t) H (x )

pSiφ (t)

Sx e φ e Sx e φ,x e

(4.24)

Sy , pSz , pSz , qSx and qSx are defined in the

113

The tension load vector for the i-th finite element is given by:

I

p (x , t) Hv,x (xe ) p (t)

Ty e

Z li Tiv

pTi (t) = pITz (xe , t) Hw,x (xe ) dxi = pTiw (t) (4.25)

0

0 p (t)

Tiφ

where the distributed tension-induced loads are defined in Eq.(2.86) in Section 2.4.5.

the elemental nodal load vector needs to be computed . The lag damper applies a

moment, MD , to the inboard end of the blade, which is applied directly to the nodal

The quantities MDx , MDy and MDz are the components of the lag damper moment

(Eq.(2.87)).

to the tip local undeformed coordinate system is given by Eq.(2.29), where the

yLt = [ΛLG ] yG

t

(4.27)

where the superscript t indicates that these quantities correspond to the tip element;

the subscripts L and G denote the local and global coordinate system, respectively.

114

The vectors of degrees of freedom in both coordinate systems are:

h iT

yLt = v1 v,x1 v2 v,x2 w1 w,x1 w2 w,x2 φ1 φ2 φ3

(4.28)

and

h iT

t

yG = vJ v,xJ vT v,xT wJ w,xJ wT w,xT φJ φM φT uJ (4.29)

The transformation matrices between the global and the local coordinate sys-

tems are nonlinear, as shown by Ref. 97. The derivation in Ref. 97 is for a blade tip

with sweep, anhedral and pretwist rotations. The result of Ref. 97 for sweep only

(as no anhedral and pretwist are used in the present study) follows.

uL u

G

vL = [TΛ ] vG (4.30)

w

w

L G

The nonlinear elements of the transformation are due to the rotational degrees

of freedom, [φG , w,xG , v,xG ]T in the global coordinate system and [φL , w,xL , v,xL ]T in

φ φG

L

T K

w,xL = [TΛ ] + TΛ w,xG (4.31)

v

v

,xL ,xG

115

The matrix TΛK contains the nonlinear elements of the transformation be-

tween the rotational degrees of freedom in both frames. For the case of sweep only,

the only nonlinear terms appear in the lag equation, and the only non-zero element

(4.32)

[Λ ] [0] [ΛN L ]

LL

[ΛLG ] = [0] [ΛF F ] [ΛF T ] (4.33)

[0] [ΛT F ] [ΛT T ]

With the nonzero submatrices given by:

cos Λ 0 0 0

0 1 0 0

[ΛLL ] =

(4.34)

0 0 cos Λ 0

0 0 0 1

1 0 0 0

0 cos Λ 0 0

[ΛF F ] =

(4.35)

0 0 1 0

0 0 0 cos Λ

0 0 0

− sin Λ 0

0

[ΛF T ] =

(4.36)

0 0 0

0 0 − sin Λ

0 sin Λ 0 0

1 1

[ΛT F ] = 0 2 sin Λ 0 2 sin Λ

(4.37)

0 0 0 sin Λ

116

cos Λ 0 0

[ΛT T ] = (4.38)

0 cos Λ 0

0 0 cos Λ

0 0 0

w,xJ sin2 Λ + sin Λ cos Λv,xJ

0 0

[ΛN L ] =

(4.39)

0 0 0

0 0 0

The transformation between the local and the global coordinate systems of the

The blade mode shapes and corresponding natural frequencies are calculated using

the rotating blade in vacuo with no damping. The problem of finding the mode

where [M ] is the linear portion of the mass matrix, [K] is the linear portion of

the stiffness matrix and yn is the vector of nodal displacements from the finite

element model. Because the mass and stiffness matrices are non-linear, they are

never built explicitly, and therefore they must be calculated using a finite difference

approximation.

The mass and stiffness matrices are constructed by successively perturbing the

117

The calculation of the blade mode shapes involves the solution of the eigen

problem

where ȳn are the eigenvectors which form the columns of the modal transformation

matrix, [V ].

The natural frequencies of vibration of the system, ω, are the square roots of

the eigenvalues. ωi is the natural frequency corresponding to the ith mode shape,

With the exception of the two lowest mode shapes, which represent the rigid

flap and lag modes for an articulated rotor, all eigenvectors contain a combination

of flap, lag and torsion. The reason for this is found in the coupled nature of the

flexible blade modes. However, in this study each mode will be referred as a flap,

tip model

If the tip sweep is modeled, the mass and stiffness matrices need to be built sepa-

The vector of nodal degrees of freedom for the blade needs to be split in two

subvectors, one containing the degrees of freedom of the straight portion of the

blade, and one for the degrees of freedom corresponding to the swept tip elements.

The degrees of freedom corresponding to the node connect the straight and swept

118

parts of the blade appear in both vectors. The global degrees of freedom for a blade

with four finite elements for the straight blade and one finite element for the swept

For a general case with Ne total final elements and Nt finite elements in the

swept tip, the assembled vector of degrees of freedom for the straight portion of the

yS = [ vS wS φS ]T (4.42)

where

T

vS = v0 v0,x v1 v1,x . . . vNe −Nt v(Ne −Nt ),x (4.43)

T

wS = w0 w0,x w1 w1,x . . . wNe −Nt w(Ne −Nt ),x (4.44)

T

φS = φ0 φ1 . . . φ2(Ne −Nt ) (4.45)

Similarly, the vector of degrees of freedom for the swept part of the blade is

given by

yT = [ vT wT φT ]T (4.46)

where

T

vT = vNe −Nt v(Ne −Nt ),x . . . vNe v(Ne ),x (4.47)

T

wT = wNe −Nt w(Ne −Nt ),x . . . wNe wNe ,x (4.48)

T

φT = φ2(Ne −Nt ) . . . φ2Ne (4.49)

119

and

The construction of the straight portion of the blade mass and stiffness ma-

trices, MS and KS respectively, is performed as in the unswept model, but with the

total number of finite elements and the number of degrees of freedom limited to

Similarly, the mass and stiffness matrices corresponding to the swept blade,

MT and KT , are built separately but in the same manner as the rest of the blade,

only limited to the number of finite elements and degrees of freedom of the tip

portion of the blade. For the swept tip element, the mass and stiffness matrices

need to be obtained in the local tip coordinate system described in Section 2.2.2.

Once calculated in the swept coordinate system, the tip mass and stiffness

for the stiffness matrix, where the substript (...)G indicates that the matrix is in

the global (blade undeformed preconed) coordinate system and (...)L refers to the

local (tip undeformed) coordinate system. The transformation matrix ΛLG is that

defined in Eq.(4.33).

120

Like for the unswept case, described in Section 4.3, the calculation of the blade

The assembly of the straight and swept portions of the mass and stiffness ma-

trices must include the reordering of the degrees of freedom (Ref. 121). A simplified

represent the mass and stiffness matrices of the straight unswept part of the blade,

and MT and KT represent the mass and stiffness matrices of the swept part of the

bade, after they have been transformed to the global coordinate system (Fig. 4.4 is

an over-simplification of the process, and the actual assembly has to separate the

yn = [V ] q (4.55)

coefficients.

121

The matrix [V ] is formed with the normal modes of the blade, columnwise.

Therefore, if Nm modes are used in the modal coordinate transformation, the matrix

The nodal load vectors for each element, defined in Eqs.(4.22) and(4.23) through(4.26)

are transformed into modal load vectors using the same modal coordinate transfor-

mation used to reduce the number of degrees of freedom. The transformations are

as follows:

Ne

X

FA = [Vi ]T pAi (4.56)

i=1

XNe

FI = [Vi ]T pIi (4.57)

i=1

XNe

FS = [Vi ]T pSi (4.58)

i=1

XNe

FT = [Vi ]T pTi (4.59)

i=1

XNe

FD = [Vi ]T pDi (4.60)

i=1

The blade dynamics are then represented by the ordinary differential equations

resulting from adding the contribution of all the modal load vectors:

0 = FA + FI + FS + FT + FD (4.61)

where the total number of modal equations is obtained by multiplying the number of

Nm .

122

4.5 Trim

The generic trim flight condition is a steady, coordinated, helical turn. The flight

condition is defined by three parameters: the velocity V along the trajectory, the

flight path angle γ, which is positive for climbing flight, and the rate of turn ψ̇,

positive for a right turn (Refs. 122, 123). The geometry of the trim problem and

Straight and level flight becomes then a particular case in which both the

flight path angle and the rate of turn are zero. In climbing and descending flight,

the flight path angle would be nonzero, while in turning flight it is the turn rate

some singularities appear in the model (in those equations in which the velocity is

the denominator). To avoid this problem, a very small value of V in the forward

X = { XB XR X I } T (4.62)

body, rotor and inflow. If the free wake is used to compute the induced velocities,

123

the inflow unknowns are not necessary, and therefore the vector of trim unknowns

becomes:

X = { XB XR }T (4.63)

used, or 9 + (1 + 2Nh )Nm if the free wake model is included, of which 9 correspond

to XB , Nm (2Nh + 1) to XR , and 3 to XI .

where θ0 , θ1c , θ1s and θ0t are the collective, cyclic and tail pitch settings, respectively,

and αF , βF , θF and φF are angle of attack, sideslip, pitch angle and bank angle of

the fuselage; λt is the constant tail rotor inflow, and is included in the xB partition

for convenience, as it is present both when dynamic inflow and the free wake models

are used.

The angle of attack αF and the sideslip βF , together with the velocity along

the trajectory V , are used to calculate the velocities u, v, and w along the body

axes

vF = V sin βF (4.66)

The angular velocities p, q, and r about the body axes can be calculated from the

124

Euler angles θF and φF and the turn rate ψ̇:

Fourier series. The Fourier expansions of the coefficients of the generalized coordi-

nate of each of the blade modes become the unknowns of the trim problem:

Nh

X

k k

q (ψ) ≈ qapp (ψ) = q0k + k

(qjc k

cos jψ + qjs sin jψ) (4.71)

j=1

where q0k is the constant coefficient in the expansion of the k-th mode, and qjc

k k

and qjs

are the coefficients of the j-th harmonic cosine and sine for the k-th mode; Nh is the

number of harmonics included in the expansion for each mode, and Nm is the number

of main rotor modes retained for the trim procedure after the modal coordinate

defined as

XR = [ q01 q1c

1 1

q1s 1

q2c 1

q2s 1

. . . qN hc

1

qN hs

...

. . . q0Nm q1c

Nm Nm Nm Nm Nm Nm T

q1s q2c q2s . . . qNhc

qNh s ] (4.72)

Assuming that the blades are identical, and perform identical motions in trim,

If a dynamic inflow model is used (Refs. 6, 36), the XI partition contains the

values of the dynamic inflow coefficients representing the main rotor inflow:

XI = [ λ0 λs λc ]T (4.73)

125

where λ0 is the uniform inflow component, and λs and λc are the sine and cosine

The trim equations make up a system of non-linear algebraic equations, which can

be written as

F(X) = 0 (4.74)

F = { FB FR FI }T (4.75)

F = { FB FR }T (4.76)

with the free wake model, where FB represents the equations of the body, FR , those

Body equations FB

A set of nine algebraic equations is used to describe the trim state of the entire

aircraft and the tail rotor inflow. These equations are the following:

Force and moment equilibrium is enforced by requiring that the linear and

angular accelerations of the aircraft be equal to zero when averaged over one

Z 2π

u̇F dψ = 0 (4.77)

0

126

and similarly for v̇F , ẇF , ṗ, q̇ and ṙ.

The condition for turn coordination is that the Y force component be zero

" #

Z 2π

ψ̇V

sin φF − (cos αF cos φF + sin αF tan θF ) cos βF dψ = 0 (4.78)

0 g

The flight path angle, γ, and angle of attack, sideslip angle, roll angle, and

Z 2π

[cos αF cos βF sin θF − (sin βF sin φF + sin αF cos βF cos φF ) cos θF

0

− sin γ] dψ = 0 (4.79)

The tail rotor equation enforces that the tail rotor inflow be constant on av-

Z 2π

ν˙t dψ = 0 (4.80)

0

Rotor equations FR

The ordinary differential equations governing the blade motion, which is assumed to

technique (Ref. 95). The generalized coordinates and their derivatives are obtained

127

from their truncated Fourier series expansions, Eq.(4.71):

Nh

X

k k

q (ψ) ≈ qapp (ψ) = q0k + k

(qjc k

cos jψ + qjs sin jψ) (4.81)

j=1

Nh

X

k k k k

q̇ (ψ) ≈ q̇app (ψ) =Ω (−qjc sin jψ + qjs cos jψ) (4.82)

j=1

Nh

X

k k 2 k k

q̈ (ψ) ≈ q̈app (ψ) = −Ω (qjc cos jψ + qjs sin jψ) (4.83)

j=1

where the superscript k refers to the k-th mode in the modal coordinate trans-

formation and Nh is the highest harmonic in the truncated Fourier series for the

T

q 1 q 2 . . . q Nm

q= (4.84)

The approximations to q and its derivatives from the Fourier series expansions

1 2 Nm

T

qapp = qapp qapp . . . qapp (4.85)

j j

, and qks that minimizes on av-

erage the residual vector ε is the one which satisfies the following equations (Ref. 122)

128

Z 2π

εk (ψ) dψ = 0

0

Z 2π

εk (ψ) cos jψ dψ = 0 j = 1, . . . , Nh (4.87)

0

Z 2π

εk (ψ) sin jψ dψ = 0 j = 1, . . . , Nh

0

2Nh )Nm scalar equations (Nm is the number of modes used in the modal coordinate

The dynamic inflow trim equations require that the average of the derivative of each

dynamic inflow coefficient over one rotor revolution be zero, that is:

Z 2π

λ̇0 dψ = 0 (4.88)

0

Z 2π

λ̇s dψ = 0 (4.89)

0

Z 2π

λ̇c dψ = 0 (4.90)

0

When used to determine the induced velocities, these three equations make

The widely used, nonlinear algebraic equation solver HYBRD (Ref. 124) is used

129

a modified Powell hybrid method (Ref. 125), which is described here briefly as a

preliminary to the description of the trim process with the free wake.

F(x) = r (4.91)

into the system of equations F. The equation solver then adjusts the values of x to

reduce the norm of r below a certain tolerance, and therefore solve the trim problem.

HYBRD begins the solution process by building a Jacobian matrix using forward

difference, that is, by perturbing each of the elements of the x vector. The Jacobian

is used to compute a direction along which the residuals are minimized, and the

residuals are recomputed along this direction until the residual stops decreasing.

Then the Jacobian is recalculated and the process is continued until convergence.

in trim

When the time-marching free wake model is included in the trim calculation, the

main rotor inflow coefficients and the dynamic inflow equations corresponding to

the main rotor are not included in the vectors of trim unknowns and trim equations

respectively. The induced velocities are instead provided by the free wake model,

In the coupling of the flight dynamics model with the free wake, the former

130

1. The hub linear and angular velocities, uGW , vGW , wGW , pGW , qGW , rGW .

The hub linear velocities, uGW , vGW , wGW , Eqs.(2.38,3.5), and angular veloc-

ities, pGW , qGW , rGW , Eq.(2.41,3.6), in the global wake coordinate system de-

scribed in Section 3.3.4, are required to calculate the velocity field at the vortex

filament, as described in Section 3.5. The total velocity at any point in the

V (r(ψ, ζ)) = V∞ + Vind (r(ψ, ζ)) + Vman (r(ψ, ζ)) (3.13) repeated

The velocity distribution over the rotor due to the maneuver, pGW , qGW and

2. The distribution of the velocities at all points in the blade around the azimuth,

not including the induced velocities from the free wake model, in all three

directions, Vx (ψ, r), Vy (ψ, r), Vz (ψ, r), Eq.(2.50), which are computed as de-

The calculation of the bound circulation with the Weissinger-L model is de-

131

scribed in Eq.(3.23) in Section 3.6:

NS

X

Ibi,j + IN W i,j Γj = {(V∞ + VBM + Vman + VF W ) · n}i (3.23) repeated

j=1

This calculation requires a knowledge of the velocities seen at the blade section,

both those of translational (free stream and blade motion) and rotational (ma-

the rotor-fuselage model and are provided to the free wake model to be used in

the computation of the circulation, contained in the velocities Vx (ψ, r), Vy (ψ, r)

and Vz (ψ, r) which are the components of the total velocity at each blade sec-

tion. This velocity contains the linear translational velocities of the helicopter

in global wake axes, uGW , vGW and wGW , the velocities due to the angular

rates, pGW , qGW and rGW , described above, and the velocities due to the blade

dRB ∂RB

= + ω × RB (2.40) repeated

dt ∂t

Note, however, that the Weissinger-L model needs these velocities at the 3/4

chord location, not at the elastic axis (located in the quarter-chord location),

which is the point at which they are required in the rotor-fuselage model

for the rest of the aerodynamic load calculation. To account for that, the

velocities due to the blade motion are computed as in Eq.(2.40), but using the

which is the position vector of a point which is not necessarily on the elastic

132

axis and where the underlined terms represent the distance of the point from

the elastic axis. This way, the velocities can be calculated at the 3/4 chord

location.

The flapping angles, β(ψ), are necessary to determine the relative positions

of the blade points with respect to the wake filaments in the flow field, which

requirement is the determination of the release point of the vortex filament that

is released at the tip of the blade. In addition, the motion of the blade affects

the sectional velocity seen at each blade point (Refs. 15, 40), Eq.(2.50), needed

for the calculation of the bound circulation. The latter requirement is met

with the inclusion of the blade motion in the velocities provided by the rotor-

fuselage model to the free wake (described earlier in this section), therefore the

velocities used for the calculation of the bound circulation account for blade

flexibility.

However, the flap angle at the tip of the blade still needs to be provided by the

flight dynamics model to the free wake model. The free wake model assumes

the blade to be straight, with a flap hinge at the axis of rotation. In the flight

dynamics model, the blades are flexible and have hinges that need not be in the

axis of rotation. For this reason, an “equivalent” blade flap angle is defined.

Figure 3.2 illustrates the definition of this flap angle β, which represents the

angle between the hub plane and a straight blade hinged at the axis of rotation

133

that has the same tip flap displacement as the elastic blade. The “equivalent”

wtip (ψ)

β(ψ) = + βp (4.92)

R

wtip is the flapping displacement of the elastic blade from the undeformed

preconed blade coordinate system in the flight dynamics model. This dis-

that vector in the z-direction of the free wake blade preconed coordinate sys-

tem at the tip of the blade. The angle βp is the blade precone angle, which

relates the hub rotating coordinate system (Section 2.2.2) to the preconed un-

deformed coordinate system (Section 2.2.2). The blade flapping angle β(ψ)

therefore includes contributions from both the blade precone angle and the

The product of the free wake is the distribution of the induced velocities over

the rotor, λ(ψ, r). The present model makes use of the z-component of these induced

velocities at the rotor plane, even though the free wake model calculates the induced

velocities at the control points on the rotor in all three directions. This assumption

can be made safely for the majority of flight conditions, since the x and y components

are much smaller than the z component and the inflow can be assumed to be uni-

directional. However, the results presented in Chapter 7 seem to indicate that, for

very extreme flight conditions such as the vortex ring state, these velocities might

The induced velocities are used in Eq.(2.47) and can then added to the ve-

134

locity of the blade section, Eq.(2.44), to obtain the total local velocities required to

For the flight dynamics model and the free wake model to interact correctly,

several other details need to be taken into account, such as proper transformation

between the coordinate systems used by both models, as described in Section 3.3.3.

The trim procedure with the free wake model is described schematically in

1. The process starts with a guess of the trim solution, X0 , which is provided to

2. For each guess, XK , the velocities seen by the blade, Vx (ψ, r), Vy (ψ, r), Vz (ψ, r),

3. The free wake model adds the inflow, λ(ψ, r) (from a previous solution, the

previous iteration or, in the case of starting without an initial wake solution,

4. With this circulation, Γ(ψ, r), the free wake geometry is calculated (Sec-

tion 3.4.1) and the corresponding updated induced velocities, λ(ψ, r), are de-

termined.

135

5. With the newly calculated induced velocities, steps 3 and 4 are repeated, until

6. The converged inflow, λ(ψ, r), is returned to the flight dynamics model, which

uses it to evaluate the equations of motion and determine the residual with

7. Steps 2 to 6 are repeated until the residual falls below a certain tolerance and

There are two important aspects in the above approach that significantly re-

1. First of all, the computational cost has been cut down by not updating the

inflow during the calls to build the Jacobian matrix. To calculate the Jacobian,

the non-linear algebraic equation solver firsts calls the function that evaluates

the right hand side of the equations of motion in Eq.(4.91), F(x), once to get

the baseline solution with which to build the forward difference approximation

of the derivatives, and then it perturbs each element of the vector of trim

order of 0.01% of the value of each element of x. The free wake model, however,

was almost insensitive to such a small perturbation of one of it’s inputs. For

that reason, the induced velocities obtained when calculating the baseline for

each Jacobian matrix are fixed for the entire process of building each derivative

136

attempted solution along the new direction determined by the Jacobian, the

free wake geometry and inflow are calculated normally, as described in Fig. 4.6.

2. The present study takes advantage of the Weissinger-L method included with

the free wake model, which offers the possibility of updating the bound vortex

strengths for each step of the free wake loop, reducing the overall cost of

The vector of trim unknowns used to obtain a trim solution, X, does not correspond

exactly with the state vector, y. However, the vectors of states and controls can be

reconstructed from the information contained in the vector of trim unknowns and

the information about the flight condition at which the trim solution is obtained.

The state vector needs to be calculated in the rotating frame, which is the

frame in which the equations of motion are formulated, and it is defined at a reference

azimuth angle, ψREF , which is the reference azimuth location of a reference blade

(the locations of all other blades can be determined from this reference azimuth

angle).

For the baseline case without the free wake mode, the state vector is as follows:

y

B

y= yI (4.93)

y

R

with the partitions referring to a rigid body part, an inflow part and a main rotor

part.

137

When the free wake is used to provide the vortex wake geometry and main

rotor inflow, the state vector does not contains the inflow part:

y

B

y= (4.94)

y

R

The vector yB contains the state variables associated with the motion of the

rigid body of the aircraft, which are independent on the reference azimuth angle at

which the state vector is defined. This rigid body part takes the form (the tail rotor

inflow is included in the rigid body portion of the state vector for convenience):

yB = [u v w p q r φF θF ψF λt ]T (4.95)

These components are determined from the rigid body part of the trim vector,

XR , Eq.(4.64), and the velocity, V , the flight path angle, γ, and the turn rate, ψ̇,

which define the flight condition for which the trim solution was calculated. The

velocity components along the body axes are obtained from the velocity V and the

The angular velocities p, q, and r about the body axes can be obtained turn

138

The pitch and roll Euler angles of the fuselage in the state vector, θF , φF ,

are the same as those in the trim vector, as is the tail rotor inflow, λt . The Euler

yaw angle, ψF , is taken arbitrarily, since it is defined with respect to the inertial

The vector yR contains modal coefficients for the individual main rotor blades

and their time derivatives, which are dependent on the azimuth angle. For a four-

y(ψi )R = [q11 q21 q31 q41 q̇11 q̇21 q̇31 q̇41 (4.96)

where qik and q̇ik are the generalized displacement and velocity coefficients of the i-th

blade for the k-th normal mode in the rotating frame at azimuth angle, ψi . They

Nh

X

qik = q0k + k

(qjc k

cos jψi + qjs sin jψi ) k = 1, 2, . . . , Nm (4.97)

j=1

Nh

X

q̇ik = Ω k

(−qjc k

sin jψi + qjs cos jψi ) k = 1, 2, . . . , Nm (4.98)

j=1

with the superscript k referring to the k-th mode in the modal coordinate transfor-

The azimuth angle of the current blade, the i-th blade, is calculated with

2π(i − 1)

ψi = ψREF + (4.99)

Nb

139

The vector yI contains the inflow coefficients for the dynamic inflow model,

which are independent of the azimuth angle at which the state vector is defined,

where the main and tail rotor controls in the control vector are the same as those in

the rigid body portion of the trim vector, Eq.(4.64). The derivatives of the controls

in the control vector are taken to be zero for trim since the controls themselves are

This section describes the procedure to integrate the equations of motion to deter-

ẏ = g(y, u; t) (4.101)

f (ẏ, y, u; t) = 0 (4.102)

140

In the case of the explicit formulation, a variable step, variable order Adams-

Bashforth ordinary differential equation solver is used. For the implicit formulation

(as seen in Ref. 126), the solver DASSL, a variable step, variable order method

must start from an initial trim solution. From this trimmed helicopter state, the

non-linear equations of motion are integrated in time for the prescribed duration of

the simulation. The result is a set of time histories of all the states of the model

following the pilot-prescribed input controls. The controls prescribed by the pilot

are included in the form of time histories of at least one of the four controls, i.e.,

the collective, longitudinal and lateral cyclic and pedal (in addition to prescribing

the controls, the time histories of one or more of the state variables themselves can

be specified).

When dynamic inflow is used to obtained the induced velocities over the main rotor,

the equations of motion include the Peters–He dynamic inflow equations (Ref. 36).

If the free wake model is used instead, these equations are removed from the model

and the induced velocities are instead provided by the free wake.

The time-marching free wake model used in the present study can be advanced

in time with arbitrary time step sizes ∆ψ, and therefore can be coupled with the

141

rotor-fuselage model at any desired point in the integration of the equations of

motion. Unlike in Ref. 6, in which a steady state relaxation free wake model was

used, there is no need to assume that each time the free wake is evaluated represents

induced velocities do not need to be converged and evaluated for the entire rotor

at each call of the wake. Instead, the time-marching free wake is only advanced

from the previous to the new time in the integration procedure, resulting in highly

The input required by the free wake model to obtained a time-accurate inflow

during the integration are the same as during trim (Section 4.5.5), i.e.

The hub linear velocities, uGW , vGW , wGW , Eqs.(2.38,3.5), and angular veloc-

ities, pGW , qGW , rGW , Eq.(2.41,3.6) are required to calculate the velocity field

V (r(ψ, ζ)) = V∞ + Vind (r(ψ, ζ)) + Vman (r(ψ, ζ)) (3.13) repeated

where

with V∞x = uGW and so on. The velocity distribution over the rotor due to

142

2. The velocities at all points in the blade, not including the inflow, in all three

directions, Vx (ψ, r), Vy (ψ, r), Vz (ψ, r), computed as described in Section 2.4.2.

These velocities are necessary for the calculation of the bound circulation with

NS

X

Ibi,j + IN W i,j Γj = {(V∞ + VBM + Vman + VF W ) · n}i (3.23) repeated

j=1

3. The equivalent rigid blade flapping angles, β(ψ), given by the following equa-

tion:

wtip (ψ)

β(ψ) = + βp (4.92) repeated

R

The free wake model returns the inflow distribution for that particular flight condi-

The main difference is that only the new updated values of the flap and the

blade velocities are required at the new azimuth position at which the free wake

must be evaluated for each blade, rather than the values over the entire rotor.

The methodology used in the present study is shown schematically in Fig. 4.7.

Although the solver used for the simulation is a variable-step solver, it is necessary

to choose a step size ∆ψ at which to synchronize the flight dynamics simulation and

the free wake, which is solved separately. The simulation is advanced in steps of

variable size chosen by the solver algorithm, but the solver stops every ∆ψ, at which

time the wake is advanced to the new position and the induced velocities updated.

The step used to advance the free wake model needs not be the same as the time step

of the simulation, ∆ψ. If the free wake azimuthal discretization, ∆ψF W , is smaller

than ∆ψ, several free wake time steps take place to advance to ∆ψ. The present

143

study, however, uses the same wake azimuthal discretization ∆ψF W and time step

∆ψF W (10 degrees), and thus only one free wake step is taken for each time step of

the simulation.

the trim vector Xtrim (Eq.(4.63)), which is calculated using the method pre-

sented in Section 4.5. The state vector in the rotating frame, at a reference

azimuth angle of zero, y(ψREF = 0), needs to be extracted from the trim

tion is also available from the trim solution: the trimmed condition induced

velocities, λ(ψ, r), the corresponding distribution of circulation over the rotor,

Γ(ψ, r), and the wake filaments geometry at this baseline flight condition.

3. With the inflow held fixed, the non-linear equations of motion can be inte-

grated in time using a variable step until it reaches the point where it syn-

chronizes with the wake, at a new time ti = ti−1 + ∆ψ. The vector with the

updated states, y, and the updated flap angles, β(ψ), and blade velocities,

Vx (ψ, r), Vy (ψ, r), Vz (ψ, r), at the new time step ti for all blades are stored.

4. The current states and controls of the helicopter, y and u, the blade flapping

distribution updated for the new time step, β(ψ), and updated blade section

velocities, Vx (ψ, r), Vy (ψ, r), Vz (ψ, r), are provided to the free wake model. To-

gether with the current distribution of induced velocities, λ(ψ, r), they are used

144

to evaluate the circulation Γ(ψ, r) at the new time in the simulation with the

5. The newly-calculated circulation, Γ(ψ, r), is used to advance in time the free

wake geometry until the new time step in the simulation, ti (Section 3.4.1). If

the size of the free wake time step ∆ψF W is the same as the simulation ∆ψ,

only one free wake step is needed. If the free wake azimuthal discretization

∆ψF W is smaller than the time step of the simulation (the simulation time

step needs to be a multiple of the free wake ∆ψ), several free wake steps take

place to reach ti . In that case, the induced velocities at each new step are used

to recalculate the circulation, and the process is repeated until the new time is

reached. The induced velocities for this new geometry, λ(ψ, r), are calculated

and stored.

6. The updated induced velocities, λ(ψ, r), are used to integrate the equations of

motion to the new time in the simulation, ti+1 . Steps 3, 4 and 5 are repeated

Both the variable step Adams-Bashforth algorithm and the DASSL method

need to evaluate the equations of motion at any azimuth angle between each time

step of the integration. The specific values of azimuth angle used are selected by

the integration algorithm itself and are not known prior to the integration. The free

wake model only provides the inflow at specific azimuth angles. For that reason, a

during the evaluations of the solver between the azimuth points in which the inflow

145

is available.

model

In the earlier stages of the present study (Refs. 128, 129), the trim solution was

obtained with a relaxation free wake model (Ref. 12) because it was thought that it

this approach was later dropped in favor of the present methodology for the benefits

obtained from using the same free wake model for both trimming and integrat-

ing the equations of motion, and, specially, because the trim formulation with the

accuracy, stability). A summary of that earlier approach with a relaxation free wake

146

φ

2

Torsion

φ1

φ

0

Tip

Root w1

v1

Lagwise

v1,x Flapwise w1,x

w0

v0

v0,x

w0,x

147

v5

v5,x

w5

w5,x

v4 φ9

v4,x

w4 φ8

w4,x

φ7

v3

v3,x φ6

w3 4

w3,x

φ5

v2

v2,x 3

Rigid w2 φ4

Sect ion v1 w2,x

v1,x φ3

w1

w1,x φ2

φ1 2

1 Finit e

Element s

Hinge -

Hub Blade Root

Figure 4.2: Blade degrees of freedom for the complete blade using four finite ele-

ments.

148

v6

v5 v6,x

v5,x w6

w5 w6,x

w5,x

φ10 φ11

v4 φ9

v4,x

w4 φ8

w4,x

φ7 5

v3

v3,x φ6

w3 4

w3,x

φ5

v2

v2,x 3

Rigid w2 φ4

Sect ion v1 w2,x

v1,x φ3

w1

w1,x φ2

φ1 2

1 Finit e

Element s

Hinge -

Hub Blade Root

Figure 4.3: Blade degrees of freedom for the complete blade using four finite elements

for the straight portion of the blade and one for the swept tip.

149

MS ··· MS

. ..

! " .

M = . ! .

"

MS · · · MS + MT MT

MT MT

KS ··· KS

. ..

! " .. .

K =

! "

KS · · · KS + KT KT

KT KT

Figure 4.4: Schematic of the assembly of the straight and swept portions of the mass

150

Horizon B

151

TRIM LOOP

Trim Guess

Tentative Aerodynamic

Solution Model

Solver Velocities seen by

the blade

Calculate

Rigid Blade Circulation with

Trim Solution Flapping Angles Weissinger-L

Model

Body Pitch and

Roll Rates

GEOMETRY

Trim Equations

Residuals Inflow Distribution

Figure 4.6: Scheme of the trim procedure with a time-marching free wake model

152

TRIMMED

SOLUTION

Start Simulation:

t=0

t = t + ∆t

of Motion at t blade velocities

FREE WAKE

Calculate

circulation

Time-marching

wake to t

NO Save inflow

Is simulation

complete?

YES

End Simulation:

t=T

Figure 4.7: Integration scheme procedure with a time-marching free wake model

153

Chapter 5

shapes

This section describes the general characteristics of the helicopter and the model

fully-articulated rotor system with 4 blades of 26.83 feet radius rotating at 27 rad/sec

or 260 RPM and a forward shaft tilt of 3 degrees. The blade airfoil section is the

SC 1095, for which aerodynamic data is extracted from look-up tables. The tip of

the blade is swept by 20 degrees over the outboard 1.90 feet of the radius, and the

present study includes results both with the swept tip and assuming a straight blade

tip. The hinge offset is 1.25 feet and the cuff extends outboard for another 3.83 feet.

The blade has a chord length of 1.75 feet and an equivalent linear twist of about

−18 degrees, although the blade is twisted only outboard of the cuff. The actual

blade twist from the root to tip is −14 degrees (without the inclusion of the swept

tip).

154

The tail rotor has a cant angle of 20 degrees, and therefore it generates rela-

tively strong couplings between longitudinal and lateral directional dynamics. It has

the same airfoil section and blade twist as the main rotor, although with a smaller

The horizontal stabilizer uses the NACA 0014 airfoil and has an area of 45

square feet and an aspect ratio of 4.6. Its incidence is adjustable by the flight control

system as a function of speed. The vertical stabilizer, which uses the NACA 0021

airfoil, has a surface area of 32.3 square feet and an aspect ratio of 1.92.

Table 5.1 summarizes the configuration of the UH-60A articulated rotor heli-

copter used in this study. Unless specified otherwise, the results in this study are for

of 5250 feet.

The majority of the UH-60 parameters and non-linear functions and data ta-

bles have been adapted from an existing UH-60 simulation model (Ref. 32). Further

details of the implementation of the fuselage, empennage and tail rotor models for

the UH-60A helicopter can be found in the GENHEL theory manual (Ref. 32). Fig-

ures 5.1, 5.2 and 5.3 show the lift, drag and moment coefficients, CL , CD and CM

respectively, as a function of angle of attack α and Mach number M for the airfoil

The results included in the present study have been generated with a blade

model that includes the five lowest blade modes, which represent the two rigid modes

in flap and lag plus the first elastic flap, lag and torsion modes.

155

5.1.1 Flight dynamics model details

For the straight blade configuration, four finite elements of equal length are used to

model the blade. When the swept tip is modeled, the blade is discretized into five

finite elements, four equidistant elements for the straight portion of the blade, and

finite elements used. Therefore, for the four-element straight blade Ne = 29 and for

the swept five-element blade model Ne = 35. With the root constraints, the root

flap, lag and torsional displacement for an articulated rotor are zero, and therefore

the total number of degrees of freedom is reduced by 3, to 26 for the straight blade

For each finite element, 8 Gaussian (therefore, not equally spaced) points are

considered. Therefore, there are 32 points along the blade at which structural,

inertial and aerodynamic loads are calculated. In the case of the swept tip blade,

The effect of the main rotor downwash on the fuselage and empennage is not

included in the present study. The model is not used because of numerical con-

vergence issues. However, the model itself is just a parametric extension developed

early in the history of the legacy model to help fit experimental data, but not based

on first principles. Therefore, the effect was turned off until a better modeling of

156

5.1.2 Free wake parameters

The free wake of each blade has been modeled with only the tip vortex, whose length

is 4 free turns downstream of the rotor. A 10-degree discretization is used for both

the time and space derivatives, unless otherwise noted. The near wake extends for

20 degrees behind the blade, and beyond it extends the tip vortex.

For the purpose of calculating the bound vortex and the circulation, the blade

This section presents the results of the free vibration analysis of the UH-60A rotor

blade, both for the straight blade configuration and for the effect of tip sweep. The

natural frequencies and mode shapes are calculated using the finite element method

as described in Sections 4.2 and 4.3, using four finite elements for the straight blade

The blade mode shapes corresponding to the six lower natural frequencies for the

straight blade configuration are shown in Figs. 5.4 and 5.5. The modes corresponding

to the two lowest natural frequencies are the rigid lag and flap mode, respectively.

The next two lowest frequencies represent the first elastic (and second overall) flap

and lag modes. The first torsion mode corresponds to the sixth lowest natural

frequency. These figures show that there is some coupling between the flap and

157

torsion modes due to the structural twist built into the blade. Note that the modes

are normalized with respect to the tip value of the highest magnitude, although

the difference between the displacements in bending and the angles in torsion are

of different magnitude; thus, the plot the second flap mode may appear of smaller

magnitude than the torsion mode, although in reality this does not determine the

The numerical results for the UH-60A presented in this study have been gen-

The influence of tip sweep on the natural frequencies is shown in Table 5.2. Results

are shown for the baseline straight four-element blade model, for a straight five-

element model with the same spanwise node distribution as the swept tip model,

and for a swept five-element blade configuration. Table 5.2 shows that increasing the

number of finite elements does not significantly affect the natural frequencies and

mode shapes for the straight blade. The inclusion of the swept tip affects mostly the

torsional frequency, which increases from 5.25 to 5.35 /rev, due to the increase in

the torsional stiffness (tennis racket effect). The inherent coupling between the flap

and lag degrees of freedom is reflected in a slight decrease of the flap modes, the first

of which goes from 1.035 to 1.022 /rev with the inclusion of a 20-degree sweep at the

blade tip. The increase in torsional frequency is observed in some of the studies in

the literature (Refs. 98, 101), while in others the opposite effect is observed (Ref. 93).

158

All these models use different aircraft with different blade properties. Therefore, the

reason for the different results is the tip sweep increases both the torsional stiffness

and the torsional inertia, and depending on which of these quantities increases more,

the torsional frequency will increase or decrease (Ref. 101). The results in the present

study present the same tendency as those in Ref. 130, in which the same helicopter

is used.

Figure 5.6 shows the variation of the lowest six natural frequencies with in-

creasing sweep angle, ranging from 0 to 25 degrees. As the sweep angle Λ increases,

the lag frequencies stay the same, the flap frequencies decrease slightly, while the

The blade mode shapes corresponding the the six lowest natural frequencies

for the swept tip configuration are presented in Figs. 5.7 and 5.8. As for the case of

the straight blade, the coupling between the flap and torsional degrees of freedom

is clearly visible, but the couplings are stronger with the swept tip, including the

second lag mode, which is significantly affected despite the corresponding natural

159

MAIN ROTOR

Number of blades 4

Radius R, ft 26.83

Blade chord c, ft 1.75

Rotational speed Ω, rad/sec 27.0

Tip speed Vtip , ft/sec 724.41

Longitudinal mast tilt iθ , deg −3.0

Airfoil section SC 1095

First airfoil section, ft 5.08

Blade precone βP , deg 0.0

Linear blade twist θT W , deg −18.0

Solidity σ 0.083

Lock number γ 5.11

Control phase shift ∆SP ,deg −9.7

FUSELAGE

Pitch inertia Iyy , lbs-ft2 38512.0

Roll inertia Ixx , lbs-ft2 4659.0

Yaw inertia Izz , lbs-ft2 36796.0

Ixz , lbs-ft2 1882.0

Horizontal tail surface area (ft2 ) 45.00

TAIL ROTOR

Number of blades 4

Radius Rtr , ft 5.5

Blade chord ctr , ft 0.81

Rotational speed Ωtr , rad/sec 124.62

Tip speed, ft/sec 685.41

Rotor shaft cant angle, deg 20.0

Table 5.1: Main parameters of the UH-60A Black Hawk helicopter configuration.

160

Mode Mode Frequency (/rev)

Number Type 4 FEM 5 FEM 5 FEM

straight straight swept

1 1st lag 0.2680 0.2679 0.2688

2 1st flap 1.0352 1.0352 1.0225

3 2nd flap 2.8160 2.8157 2.7684

4 2nd lag 4.6504 4.6551 4.6543

5 3rd flap 5.1744 5.1792 5.1073

6 1st torsion 5.2486 5.2531 5.3534

161

1.5

1.0

M=0.0

M=0.2

CL M=0.4

0.5

M=0.6

M=0.8

M=1.0

0.0

−0.5

−5 0 5 10 15 20

Angle of attack, α (deg)

Figure 5.1: Lift coefficient CL as a function of the angle of attack α and Mach

0.6

M=0.0

M=0.2

M=0.4

M=0.6

M=0.8

M=1.0

0.4

CD

0.2

0.0

−5 0 5 10 15 20

Angle of attack, α (deg)

Figure 5.2: Drag coefficient CD as a function of the angle of attack α and Mach

162

0.1

M=0.0

M=0.2

M=0.4

M=0.6

M=0.8

M=1.0

0.0

CM

−0.1

−0.2

−5 0 5 10 15 20

Angle of attack, α (deg)

Figure 5.3: Moment coefficient CM as a function of the angle of attack α and Mach

163

frequency = 0.268 /rev

1.0

Non−dimensional

displacement 0.5

0.0

(a) First Lag

−0.5

1.0

Non−dimensional

displacement

0.5

0.0

(b) First Flap

−0.5

1.0

Flap

Non−dimensional

Lag

displacement

0.5 Torsion

0.0

−0.5

0.0 0.2 0.4 0.6 0.8 1.0

Non−dimensional spanwise coordinate

Figure 5.4: First, second and third natural mode shapes for the baseline model

164

frequency = 4.65 /rev

1.0

Non−dimensional

displacement 0.5

0.0

(a) Second Lag

−0.5

1.0

Non−dimensional

displacement

0.5

0.0

(b) Third Flap

−0.5

1.0

Flap

Non−dimensional

Lag

displacement

0.5 Torsion

0.0

−0.5

0.0 0.2 0.4 0.6 0.8 1.0

Non−dimensional spanwise coordinate

Figure 5.5: Fourth, fifth and sixth natural mode shapes for the baseline model

165

7

1st lag 2nd lag

1st flap 3rd flap

6

2nd flap 1st torsion

5

Frequency (/rev)

0

0 5 10 15 20 25 30

Figure 5.6: First six natural frequencies with varying tip sweep angle.

166

frequency = 0.269 /rev

1.0

Non−dimensional

displacement 0.5

0.0

(a) First Lag

−0.5

1.0

Non−dimensional

displacement

0.5

0.0

(b) First Flap

−0.5

1.0

Flap

Non−dimensional

Lag

displacement

0.5 Torsion

0.0

−0.5

0.0 0.2 0.4 0.6 0.8 1.0

Non−dimensional spanwise coordinate

Figure 5.7: First, second and third natural mode shapes for the model with

quency=2.768/rev (bottom).

167

frequency = 4.654 /rev

1.0

Non−dimensional

0.5

displacement

0.0

−0.5

(a) Second Lag

−1.0

1.0

Non−dimensional

displacement

0.5

0.0

(b) Third Flap

−0.5

1.0

Flap

Non−dimensional

Lag

displacement

0.5 Torsion

0.0

−0.5

0.0 0.2 0.4 0.6 0.8 1.0

Non−dimensional spanwise coordinate

Figure 5.8: Fourth, fifth and sixth natural mode shapes for the model with

quency=5.353/rev (bottom).

168

Chapter 6

Trim results

6.1 Overview

This chapter describes the trim results for a variety of flight conditions. The first

section presents trim results obtained at different speeds in straight and level flight.

The simulation results are also compared to flight test data. Both the trim results

and the blade dynamics are studied, as well as the rotor loading and the effect that

the free wake inflow prediction has on it. The effect of blade flexibility, the level

of discretization of the free wake geometry and the inclusion of the swept tip are

discussed. The second section contains results for coordinated turns, with different

values of the turn rate, both to the right and the left. Special attention is paid to the

influence of the wake geometry and the induced velocities it produces on the rotor.

The third section presents results for straight climbing and descending flight. For a

range of flight path angles, the behavior of both the fuselage and rotor is studied.

Additionally, some high rate of descent conditions are to determine the highest rate

of descent in incipient Vortex Ring State for which a trim solution can be obtained.

The fourth section presents results for coordinated turns, including steep descents.

169

6.2 Straight and level flight

This section presents a comparison between the trim results and the flight test data

from Ref. 10, which was obtained with a gross weight of 16,000 lbs at an altitude of

5,250 ft.

Figure 6.1 shows the power required by the main rotor, in horsepower (hp), as a

function of speed. The values predicted by the model show excellent correlation with

the experimental values for most of the flight speed range considered. The power is

underpredicted below 30 kts (data below 20 kts was not available), although some

uncertainty in the accuracy of the flight test data also plays a role at those speeds.

Above 100 kts, the main rotor power required is slightly overpredicted, between 10%

The collective stick value is shown in Fig. 6.2, and shows a very similar type

The longitudinal cyclic stick position, presented in Fig. 6.3, has a good agree-

ment with experimental values above 60 kts. At lower speeds, the predictions are less

accurate, and the trend with speed is in the opposite direction than in the flight test

data. This discrepancy might be caused by the fact that the effect of the downwash

of the rotor on fuselage and empennage is not included in the model. At low speed

this effect can significantly alter the predicted orientation of the helicopter fuselage

170

necessary to accurately predict the longitudinal cyclic at low speed.

The fuselage pitch attitude as a function of speed is shown in Fig. 6.4. The

values predicted are good at high speed flight, but overpredicted at low speed. This

interaction model.

Figure 6.5 shows a comparison of the lateral cyclic stick position predictions

with the experimental results. Except for the two data points at around 30 kts, the

Figure 6.6 shows the correlation for the fuselage bank angle. The agreement

is good for flight speeds between 30 and 100 kts. Below 30 kts, the comparison

is inconclusive because of the scatter of the flight test data, but the trend seems

correct. At high speed flight, the roll attitude is underpredicted, perhaps because

the power overprediction results in an increased tail rotor thurst and slightly more

Finally, the pedal correlation is shown in Fig. 6.7. The results obtained with

the model show a consistent overprediction of about 10%. This overprediction of the

pedal setting has been observed in other studies (Refs. 10, 6). Ref. 6 uses a similar

rotor-fuselage model with both a relaxation free wake model (Ref. 28) and the Pitt–

Peters dynamic inflow model (Ref. 24) to simulate the same flight conditions. Both

inflow models predict very similar results, which indicates that the reasons for the

discrepancy with the experimental data are not to be found in the inflow model,

but elsewhere, possibly in the modeling of the body of the helicopter (fuselage and

empennage).

171

The values of the pitch and roll attitudes predicted with the time marching

free wake model are also similar to those obtained with dynamic inflow, as shown in

Ref. 6. Again, any discrepancies with the flight test data are unlikely to be related

This section presents the trim results obtained at speeds of V = 1, 40 and 100 kts,

hover, low speed and high speed conditions. In the first two cases there is consid-

erable mutual interaction between rotor and wake, and between the vortices trailed

by the four blades. These interactions are less strong in the third case, where the

The side view of the wake geometry as a function of speed is shown in Fig. 6.8.

In hover, the wake presents a helical structure that only breaks far downstream of

the rotor due to the instability of the wake vortices in hover, a behavior that has

been observed both experimentally (Refs. 131–133) and numerically (Refs. 134, 7).

convects downstream of the rotor. As the speed increases, the wake begins to skew,

and the vortices begin to interact as their proximity increases. The rollup of the

lateral edges of the wake geometry that characterizes the wake in forward flight, due

to the formation of vortex bundles or ”super vortices” (Ref. 15), becomes visible at

172

V = 40 kts. At V = 100 kts, the wake is convected fast behind the rotor, and the

influence of the vortices at the rotor plane is much reduced. The wake geometry is

highly skewed at this speed, and the filaments are quickly left behind, decreasing

Figure 6.9 shows the rear view of the wake geometry at V = 1, 40 and 100

kts. In hover, the rear view is very similar to the view from the side, with the

same helical pattern observed. The helicopter in hover has a negative roll attitude

the rollup of the wake on the lateral edges becomes apparent, particularly on the

advancing side. This effect becomes more pronounced as the speed increases.

The wake view from the top is shown in Fig. 6.10. In hover, the filaments

appear nearly concentric with filament age. At V = 40 kts, the wake is convected

behind the rotor, but the tip vortices still remain close. The rollup on both sides of

the wake is also visible from the top. At V = 100 kts, the tip vortices are convected

The induced velocity as a function of speed is shown in Fig. 6.11. Figure 6.11(a)

shows the inflow distribution obtained with the time-marching free wake model for

V = 1 kt. In near hover conditions, the inflow distribution looks almost axisym-

metric. The symmetry is not perfect because the tip path plane is slightly tilted

forward and some cyclic pitch is necessary to compensate for the tail rotor thrust.

6.9(a) and 6.10(a). The norm of the inflow, kλk, at this flight condition is 0.0803

(58.20 ft/sec).

173

As the speed increases, the wake begins to skew, and the vortices begin to in-

teract between themselves as their proximity to each other increases. Figure 6.11(b)

presents the induced velocities at a speed of 40 kts. The overall induced velocities

are lower than in hover, with kλk = 0.0534 (38.65 ft/sec). The geometry of the free

wake at this speed, shown in Figs. 6.8(b), 6.9(b) and 6.10(b), has a great influence

in this distribution of induced velocities. The wake filaments start to convect behind

the rotor plane, and the area where they exert their greatest influence shifts towards

the rear of the rotor. The rollup that characterizes the wake in forward flight is al-

ready present at this speed, and affects the inflow, which has a higher downwash

in the tip region in the advancing and retreating side, close to the formation of the

Figure 6.11(c) shows the induced velocities at 100 kts. The corresponding

geometry is shown in Figs. 6.8(c), 6.9(c) and 6.10(c). At this speed, the wake is

convected fast behind the rotor, and the influence of the vortices at the rotor plane is

much reduced. A region of upwash flow develops in the front of the rotor, while the

rear of the rotor, closer to the wake trailed behind, contains the larger downwash.

The overall norm of the induced velocities at this speed is even farther reduced in

comparison with the inflow at V = 40 kts, with kλk = 0.0269 (19.48 ft/sec), as the

ponent of the airflow at the section, UP , and therefore decrease the angle of attack.

Figure 6.12(a) show the distribution of angle of attack for V = 1 kt. At near hover

174

conditions, when the induced velocities are almost uniformly distributed and at their

highest value, the aerodynamic angle of attack α, which includes both the geometric

pitch (which is the sum of the pitch due to the controls, the built in twist and the

elastic twist, as seen in Eq.(2.55) and the induced angle of attack, is also almost

uniform with a small variation between 3 and 5 degrees. Because of the 3-degree

forward tilt of the rotor and the azimuthal and radial variations in the geometric

pitch θG , the distribution is slightly biased and higher in the front of the rotor than

in the rear, while the angle of attack presents the opposite distribution. The lower

induced velocities seen in the front of the rotor at 40 kts (Fig. 6.11(b)) translate

into a region of higher angle of attack, seen in Fig. 6.12(b). The areas of higher

inflow, similarly, translate into lower values of the angle of attack there. The region

of upward flow in the front of the rotor produces some areas of higher angle of attack

in the front and retreating side, while the higher induced velocities in the rear of the

rotor reduce the angle of attack seen in the rear and advancing sides (Fig. 6.12(c)).

The relation between the angle of attack and the induced inflow is, however,

not as simple, as other factors come into play, such as the perpendicular component

of the airstream that results from the orientation of the rotor or change in the

tangential component UT , which has a sinusoidal variation over the azimuth with

increasing magnitude as forward speed increases, and the changes in the elastic

deformation of the blade around the azimuth (Fig. 6.26). However, it serves as an

indication of how the changes in the inflow modeling affect the flight dynamics of

the helicopter.

Figure 6.13 shows the distribution of lift coefficient, CL , over the rotor for

175

all three speeds considered, V = 1 kt, 40 kts and 100 kts. The lift coefficient is a

function of both the angle of attack, α, and the Mach number, M (Fig. 5.1). In

hover, the Mach number varies radially in the same way over the whole azimuth.

Therefore, the only variation around the azimuth for a given radial location in the

CL distribution in hover is due to the changes in the angle of attack. Figure 6.13(a)

shows the lift coefficient at V = 1 kt, which is higher in the same areas of the rotor

where the angle of attack is higher, and similarly lower in the areas where α is lower.

As the speed increases, the effect of the Mach number on CL should be accounted,

as the changes are not solely due to variations in the angle of attack distribution.

However, the influence of the angle of attack α on the lift coefficient CL is still very

direct. Figure 6.13(b) shows the lift coefficient distribution at V = 40 ks. There

are two areas of higher lift coefficient, with values of CL up to 0.75, in the front-

retreating side (between ψ = 200◦ and ψ = 270◦ ), and at the rear of the rotor disk

(between ψ = 340◦ and ψ = 20◦ ), which occur precisely in the regions where the

angle of attack α is the highest, reaching values up to 7 to 8 degrees. The same can

be observed at V = 100 kts, shown in Fig. 6.13(c), speed at which the higher areas

values of CL of 1.

The square of the Mach number, M , increases radially outwards, uniformly around

the azimuth in hover and increasingly more biased to the advancing side as the speed

due to the influence of the square of the Mach number, which increases radially

176

outboard of the blade in the same proportion in all directions in hover. A close

examination (the scale used in Fig. 6.14(a) was chosen to be consistent with all the

speeds represented in Fig. 6.14) reveals that the influence of the angle of attack that

was clearly visible on the lift coefficient CL (Fig. 6.13(a)) is still present, although

increases in the same regions where the angle of attack α and thus the lift coefficient

CL increase, as seen in Fig. 6.14(b), and decreases similarly where the angle of attack

decreases. However, these regions are a bit more biased towards the tip, and a new

region of higher values appears in the advancing side, both effects resulting from the

higher values of the square of the Mach number towards the tips and particularly

in the advancing side (where M nears values of 0.7 at this speed). At V = 100 kts,

the elemental lift distribution (Fig. 6.14(c)), like at 40 kts, varies as a function of

the angle of attack, α. The effect of the square of the Mach number M , however,

towards slightly towards the advancing side, where M 2 is the highest. At this speed,

spanwise gradients can be observed, particularly on the second quadrant of the rotor.

These gradients indicate that the single vortex model might not be the best way

inflow is necessary (for aeroacoustic purposes) but in the case where only integral

loads are necessary (for helicopter dynamics) this assumption is still valid.

Figure 6.15 shows the local flap moment distribution, rCL M 2 , over the rotor

disk, for the cases of V = 1 kt, V = 40 kts and V = 100 kts. This value serves

as an indicative of the influence of the lift at a given point on the blade on the

177

flapping motion. At V = 1 kt, the factor of the non-dimensional radius r, between 0

and 1, reduces the value of CL M 2 inboard of the blade, while increasing it towards

the tip, where the highest contribution to the flapping motion occurs (due to the

larger moment arm). The values may appear at first sight axisymmetric, although a

slight bias towards the rear of the rotor is present. As the speed increases, the same

effect is appreciated. The lower moment arm inboard of the blade (lower value of

r) reduces the values of CL M 2 both at V = 40 kts, Fig. 6.15(b), and V = 100 kts,

Fig. 6.15(c), which therefore become biased towards the tip of the blade, however

the areas where these higher values occur are the same as for CL M 2 .

The drag coefficient, CD , like the lift coefficient CL , is also a function of angle of

attack, α, and Mach number, M (Fig. 5.2). For low Mach numbers, up to M = 0.6,

increases for all angles of attack for M > 0.6 and for angles of attack beyond α = 5◦

otherwise. Figure 6.16 shows the distribution of drag coefficient over the rotor disk

for V = 1, 40 and 100 kts. In hover, Fig. 6.16(a), the drag coefficient is almost

axisymmetric, almost constant for the inboard 3/4 of the blade, and increasing

towards the tip, where the angle of attack is higher. A small region of higher drag

occurs in the back of the rotor about the 0.5R where the angle of attack also is

higher. However, closer to the tip, the drag is higher near the front of the rotor. At

V = 40 kts, shown in Fig. 6.16(b), two areas of higher drag are observed, coincident

with the areas of higher angle of attack in the front-retreating side and at the rear

of the rotor. At high speed, Fig. 6.16(c), the high speeds reached near the tip in the

advancing side bring the blades into the transonic flow region, and compressibility

178

effects translate into areas of very high drag values.

Fig. 6.17(a), the factor of the square of the Mach number, M , makes the distri-

bution slightly more axisymmetric. The local drag is, in general, fairly low at this

speed. At V = 40 kts, Fig. 6.17(b), the values of the local drag are still low, with

areas slightly larger where the angle of attack, α, and thus the drag coefficient, CD ,

are higher. At 100 kts, the Mach number at the tip on the advancing side exceeds

M = 0.75. The local elemental drag, shown in Fig. 6.17(c), clearly shows the effect

of high Mach number and high compressibility drag caused by this same high M ,

The variations in angle of attack affect, therefore, both the induced and profile

torque generated by the rotor. Figures 6.18 and 6.19 show the distributions of

elemental induced torque, rCL M 2 sin φ, and elemental profile torque, rCD M 2 cos φ

for V = 1 kt, V = 40 kts and V = 100 kts. In near hover conditions, both

the induced and profile torque are almost axisymmetric, as shown in Figs. 6.18(a)

and 6.19(a). The slight asymmetry is primarily caused by the asymmetry of the

distribution of φ = tan−1 ( UUPT ), which in turn depends on the slight forward speed

of 1 kt and the longitudinal flapping β1c = 3 degrees. The induced torque reaches

values of 0.02 towards the tip for almost the entire azimuth, which combined with

moderately high values of profile torque rear the tip for the entire azimuth, with

values up to 0.008, contribute to high power required to hover, around 1724 hp, as

seen in Fig. 6.1. At V = 40 kts, the overall induced torque goes down, Fig. 6.18(b),

as the induced velocities are lower. Moreover, the higher values of local induced

179

torque move to the rear half of the rotor, where the induced velocities are higher.

At this speed, the profile torque is still much lower than the induced torque. Both

the induced and profile torque being lower than in hover, it follows that the power

require to fly at 40 kts is also lower, around 1133 hp. As the speed increases, the

Mach number on the advancing side increases. At V = 100 kts, the local profile

torque distribution, in Fig. 6.19(c), shows an area of very high values near the tip

of the advancing side, reaching values of 0.02 at the very tip, while it remains lower

for the rest of the rotor. The induced torque continues to decrease as the induced

velocities become lower with the highly skewed wake vortices, with maximum values

of around 0.015 over a smaller region of the rotor than at lower speeds. The increase

in torque due to the higher drag encountered at higher speeds in the advancing side

increases the overall power required, while the decrease in induced torque reduces

the power requirements, with values around 1063 hp, as shown in Fig. 6.1. As the

speed continues to increase, the profile torque becomes dominant and the power

Figures 6.20 and 6.21 present the distribution of sectional pitching moment,

kts and V = 100 kts. In hover, both CM and CM M 2 appear almost axisymmetric,

although it takes slightly higher values, even positive, at the front of the rotor,

particularly towards the tips. At V = 40 kts, the highest, positive values occur

at the rear of the rotor, although two other regions of high, albeit negative, CM

occur around ψ = 100◦ and ψ = 250◦ . The case at V = 100 kts is characterized by

a region of low, nose-down pitching moments throughout the advancing side, and

180

higher values in the front-retreating side. The variation in the pitching moment can

be related to the changes in the lift distribution. The regions where the maximum

pitching moment occurs are also those where the local lift coefficient is greater

(Figs. 6.14(a) through 6.14(c)). Since the lift coefficient shown is located at the

quarter chord, an increase in lift produces a nose up pitching moment, and similarly

lower lift is reflected in more negative or nose down pitching moment over the elastic

axis.

This section presents an analysis of the blade dynamics during straight and level

flight, for the baseline model with 4 finite elements and a straight blade configura-

tion.

Figure 6.22 shows the variation of the equivalent flapping angle at the tip of

the blade β as a function of the azimuth angle ψ at V = 1 kt, 40 kts and 100

kts. In hover, the induced velocities and angle of attack distributions are almost

axisymmetric (Figs. 6.11(a) and 6.12(a)), with slightly higher values of the angle of

attack in the rear of the rotor, as described in Section 6.2.2. A close examination of

the lift distribution (Fig. 6.14(a)) and the contribution of lift to the flapping moment,

rCL M 2 (Fig. 6.15(a)), as previously explained, reveals that the lift distribution is

slightly biased, with values moderately larger in the rear of the rotor, producing a

sinusoidal flapping response, rather than simple coning only. Figure 6.23(a) shows

that the magnitude of the higher harmonics is really insignificant compared to the

181

first harmonic. The phase of the flapping response in hover is 35 degrees, as shown

in Fig. 6.23(b). The presence of some longitudinal and lateral flapping (β1c = 1.5◦ ,

β1s = 1.1◦ ) is necessary to orient the rotor for equilibrium with the fuselage, which

is slightly oriented with a pitch attitude of θ = 4.7◦ and roll attitude of φ = −2.8◦

(Figs. 6.4 and 6.6). The longitudinal and lateral flapping coefficients respond to the

cyclic controls in Fig. 6.3 and 6.5: a longitudinal cyclic of θ1s = −1.7◦ producing a

forward longitudinal tilt of the rotor by 1.5◦ and a lateral cyclic of θ1c = 0.7◦ causing

Fig. 6.22 decreases, as fewer collective is necessary (Fig. 6.2), translating into a

reduced coning angle (β0 = 3.1◦ down from β0 = 3.5◦ in hover) associated with the

lower collective setting necessary for this flight condition (Fig. 6.2). The response

over the azimuth, unlike the case of hover, indicates the presence of higher harmon-

ics. Figure 6.23(a) shows that, while considerably smaller than the first harmonic,

the magnitude of the second and third harmonic is significant. The cause for these

higher harmonics can be found in the distribution of rCL M 2 (Fig. 6.15(b)), the con-

tribution of lift to the flapping moment has multiple regions where rCL M 2 is higher,

producing a 2/rev and a 3/rev variation in the flapping response. Figure 6.23(b)

shows that the phase of the first and principal harmonic corresponds to the region

of higher angle of attack around ψ = 0◦ , while the second and third harmonics,

of much lower magnitude, are excited by the areas of higher rCL M 2 found in the

The flapping response at V = 100 kts, depicted in Figs. 6.22 and 6.23 as

182

well, shows a decrease in the overall magnitude of the flapping angle (β0 = 2.7◦ ),

with respect to the case of V = 40 kts. However, at this higher speed the cause

is not the lower collective, which, after decreasing to a minimum at about 70 kts,

starts to increase again; instead, it is the overall lower magnitude of the angle of

attack over the rotor (Fig. 6.12(c)), which as shown in Section 6.2.2 produces lower

magnitude rCL M 2 (Fig. 6.15(c)). Some signs of the presence of higher harmonics

are also visible in the flapping response at this speed. The dominant first harmonic

is excited by the area of higher magnitude rCL M 2 found in the rear of the rotor,

while 2/rev and 3/rev variation in rCL M 2 , of smaller magnitude, excite the higher

harmonics.

The equivalent lag angle at the blade tip, ζ, is shown in Fig. 6.24, for the cases

of V = 1, 40 and 100 kts, as a function of the azimuth angle ψ. In hover, the lag

angle is almost constant around ζ = −6 degrees (ζ is positive for angles in the lead

the rotor azimuth. The higher drag seen in the front of the rotor towards the tip,

shown in Figs. 6.16(a) and 6.17(a), slightly increases the lag between ψ = 90◦ and

ψ = 270◦ , while the reduction in the drag force in the rear of the rotor allows the

blade to lag less. At 40 kts, the overall drag over the rotor, shown in Figs. 6.16(b)

and 6.17(b), decreases, and therefore the mean lag over the azimuth also decreases in

magnitude, with an average value of around ζ = −4.2 degrees, as seen in Fig. 6.24.

higher drag occurs in the rear and advancing side of the rotor, while in the front and

retreating side the drag is lower. As a consequence, the lag displacement increases

183

(becomes more negative) on the advancing side, while it decreases on the retreating

side, with an overall variation over the azimuth of half a degree. At V = 100 kts,

the drag increases considerably near the tip of the blade on the advancing side,

although it decreases over the rest of the rotor (Figs. 6.16(c) and 6.17(c)). The

overall reduction in drag reduces the average lag over the azimuth, Fig. 6.24, which

at this speed has a mean around ζ = −4 degrees, with an overall variation over

the azimuth of half a degree. The higher drag on the advancing side increases the

lag displacement, which reaches its maximum near the front of the rotor, while it

decreases in the retreating side where the lower drag forces are found.

Figure 6.25 shows the magnitude and phase of the first three harmonics of the

lag response at V = 1 kt, 40 kts and 100 kts. In all three cases, the presence of

higher harmonics of similar magnitude indicate that the drag distribution excites a

1/rev, 2/rev and 3/rev variation in the lag displacement. In hover, the three first

harmonics are similar in magnitude, although the constant lag coefficient (ζ0 = −5.9

degrees) is dominant and therefore the higher harmonics (ζ1c = 0.1, ζ1s = 0.1 and

even lower second and third harmonic coefficients) are not clearly appreciated in

Fig. 6.24. At V = 40 and 100 kts, the constant lag coefficient ζ0 is lower than in

hover, and the first harmonic is comparably larger, therefore the sinusoidal variation

The total elastic torsion deflection at the tip of the blade as a function of the

azimuth is shown in Fig. 6.26, for speeds of V = 1 kt, 40 kts and 100 kts. The

down amplitude, with an overall twist variation over the rotor of around ∆φ = 0.5

184

degrees, with a mean value of around φ = −0.5 degrees. At the V = 40 kt case, the

negative for the entire range of the azimuth, with a variation in φ between −0.7

and 1.1 degrees. Moreover, the presence of higher harmonic excitations becomes

evident. The increase in the magnitude of the peak-to-valley variation of the elastic

twist continues with the V = 100 kt case, which now ranges between −1.2 and 1.0

degree. At this speed, the elastic twist also shows signs of the presence of higher

Figure 6.27 shows the magnitude and phase of the harmonics of the elastic

torsion response at V = 1 kt, 40 kts and 100 kts. At V = 1 kt, the magnitude

of the first harmonic is clearly dominant, with a value of 0.24 degrees over the

almost negligible 0.03 and 0.02 of the second and third harmonic. At 40 kts, the

magnitude of the first harmonic increases to 0.61 degrees. The second and third

harmonic become significant, with magnitudes of 0.24 and 0.38 degrees respectively,

i.e., approximately half the value of the first harmonic. At 100 kts, the magnitude of

the first harmonic increases further to 0.97 degrees. The second and third harmonics

have magnitudes comparable to those at 40 kts, 0.37 and 0.23 degrees respectively,

but they are proportionally lower than the first harmonic than at V = 40 kts.

To understand the behavior of the torsion of the blade with speed, one must

look at the distribution of pitching moment over the rotor, shown in Fig. 6.20, or

are slightly larger at the front of the rotor than at the rear, which corresponds with

185

the simple low-magnitude, 1-harmonic twist distribution observed at this speed.

At 40 kts, the pitching moment distribution shows three identifiable areas of low,

nose-down values, which twist the blade nose-down, and three areas of higher, even

Fig. 6.26, in which the torsional deflection reaches a local minimum for each min-

imum of the pitching moment in Figs. 6.20(b) and 6.21(b), around ψ = 80◦ , 180◦

and 300◦ , while the positive areas of the pitching moment, the highest at around

ψ = 0◦ and others around ψ = 130◦ and 250◦ , produce the positive values of twist.

A similar behavior is observed at 100 kts. On the advancing side, the pitching mo-

ments are low and produce a nose-down twist of the blade, as seen in Figs. 6.20(c)

ψ = 100◦ , depicted in Fig. 6.26. The maximum values of the pitching moment are

found on the retreating side, twisting the blade up to φ = 1 degree at the rear of

the rotor.

This section studies the effect of modeling blade flexibility on the trim solution and

compares the results obtained with the flexible blade model used throughout this

thesis, which includes the five lowest natural frequencies and mode shapes, with a

simpler rigid model, which only includes the two lowest frequencies, corresponding

The predictions of main rotor power obtained using the two models are shown

186

in Fig. 6.28. The predictions of the simplified, rigid blade model is on average about

10% lower than those of the flexible blade model. In the lower speed range, this

makes the rigid blade model under-predict the power required by the main rotor. In

high speed, however, the flexible blade model over-predicts the experimental values

Figure 6.29 shows the main rotor collective setting. As for the case of the power

required, the rigid blade model predicts lower values of the collective by about 5%.

The differences in the predicted values for power and collective can be ex-

plained by looking both at the blade elastic torsion and the inflow distribution.

The elastic torsional tip displacement, shown in Fig. 6.26 indicates that with

the elastic blade, an overall nose-down twist is present on the blade. This negative

torsion reduces the geometric angle of attack of the blade. Therefore, to produce

the same thrust, the rigid blade model needs less collective.

The induced velocities predicted by the rigid blade model are lower than with

the flexible blade model, as seen in Fig. 6.30 for the case of near hover conditions.

two parameters that are affected by the modeling of the blade, namely the flapping

angle and the bound circulation. This lower flap angle is passed on the the free wake

The time histories of the flapping angle at V = 1 kt, predicted using the two

models, are shown in Fig. 6.31. The flap angle is smaller with the rigid blade model

If the elastic displacements are ignored, then the local airflow velocities will

187

be affected, and so will be the circulation and the inflow.

Figure 6.32 shows the x and z-components of the blade sectional velocity in the

blade undeformed coordinate system along the blade span at ψ = 0 and ψ = 180

degrees at V = 1 kt. The plots show that the differences between the rigid and

flexible blade models are significant, The bound circulation, which is calculated

with the local airflow velocities (see Section 3.6), will therefore be also affected.

and ψ = 180 degrees. Except for the root of the blade, the flexible blade model pre-

dicts an overall higher bound circulation, which translates into stronger tip vortices

This section presents the results of including the swept tip in the blade modeling.

Figure 6.34 shows the comparison of the main rotor power required with experi-

mental values, both with the simplified straight blade and the swept tip blade, as a

function of flight speed. For V less than 40 kts, the power required with the swept

tip blade is lower than the equivalent straight blade configuration. Between 40 and

100 kts, the power predicted by the swept tip blade model is higher. For V greater

than 100 kts, it becomes progressively lower. The difference in predictions can be

The collective setting as a function of speed, both with the straight and swept

blade models, is shown in Fig. 6.35. At very low and very high speeds, both models

188

trim to a similar collective position. In the middle range of speeds, however, the

swept tip blade model trims to higher values of collective, significantly overpredicting

the experimental values. These higher collective angles increase the total geometric

angle of attack and therefore affect the flapping response, the bound circulation and

the overall induced velocities, and therefore, the power required. This higher values

of collective are also seen in other studies in the literature, such as in Ref. 130.

The elemental induced and profile torque distributions at hover, 60 kts and

158 kts are shown in Figs. 6.36 through 6.37 (note that a different color range is

used for the profile power at 158 kts because the values at that speed are a full

order of magnitude higher than in the other speeds shown, and using the same

color scheme for all would render the plots indistinguishable at the lower speeds),

with a straight tip blade model (top) and with the swept blade model (bottom).

Figure 6.37(b) shows that, at V = 1 kt, the elemental profile torque inboard of the

swept tip junction is higher than in the corresponding straight blade model, but

perpendicular to the blade section. The elemental induced torque, rCL M 2 sin φ,

shown in Fig. 6.36(b), is very similar both with the straight and the swept tip

blade models, although the swept blade model shows slightly higher values in the

retreating side. The overall effect is that the power required to hover is slightly

lower for the swept tip model. At 60 kts, the elemental profile torque for the swept

tip blade is significantly lower, compared to the straight blade (Fig. 6.37(d)). This

reduction, however, does not make up for the higher induced torque (Fig. 6.36(d)),

which increases specially in the rear of the rotor. As a result, the overall power with

189

the swept tip is higher. At V = 158 kts, both the overall induced and profile torque

distributions result in a lower power required for the swept tip blade, which is in a

Figures 6.38(b) through 6.38(f) show the distributions of induced velocities for

V = 1 kt, 60 kts and 158 kts with a straight tip blade model and with the swept

blade model. In hover, the induced velocities predicted by both blade models are

very similar, although the swept tip model is slightly higher (the overall inflow norm

is kλk = 0.0804 with the straight tip, and kλk = 0.0822 with the inclusion of the

swept tip). At V = 60 kts, the swept tip model produces higher induced velocities

in the rear of the rotor, which increases the overall inflow (from kλk = 0.0373 for

the straight blade model to kλk = 0.0464 for the swept one). At V = 158 kts, the

inflow over most of the rotor is lower with the swept tip model, although there is a

The bound circulation distributions for the same cases above is shown in

Figs. 6.39(b) through 6.39(f). In hover, the circulations obtained are very simi-

lar, which explains the similar induced velocities observed earlier. At V = 60 kts,

the swept tip blade has regions of higher induced velocity, particularly on the re-

treating side. This translates into a higher tip vortex strength, which explains the

higher induced velocities seen in that region (Fig. 6.38(d)). At V = 158 kts, there

is a region of very high circulation in the front of the rotor, which produces a strong

tip vortex that in turn causes the very high values of inflow close to the tip observed

in Fig. 6.38(f). Notice that the bound circulation at this high speed does now show

signs of strong tip vortex on the advancing side. This may be an indication at at

190

these high speeds the single-tip vortex model might not be sufficient and multiple-

trailed vortices might be necessary. The present study, however, will not address

such aerodynamic issues, and the baseline wake model with a single tip vortex will

be used throughout.

Notice that to include the effect of sweeping the blade tip in the free wake

model, the only approach consisted of moving the collocation points in the blade

the straight and swept elastic axes at each given radial location. No further con-

sideration has been made in the free wake to account for the swept tip. Given that

there have not been any particular studies of the stand-alone free wake model with

blades with swept tips, this was the only approach considered. A recommendation

for the future would include the study of the free wake with swept tip blades to bet-

ter understand the behavior seen in the present study, which might include multiple

Several turns are performed with a velocity along the trajectory of V = 60 kts and

The experimental trim results (Ref. 10) are presented as a function of the roll atti-

tude rather than the turn rate, therefore, the same representation will be used for

191

all the results presented in this section. The relation between the roll angle and the

turn rate is given in Fig. 6.40. This relation is almost linear for the range of turn

rates considered in this study. Therefore, for the sake of simplicity, both terms will

A parameter that helps indicate the severity of the turn is the load factor. For

the cases in the present study the values are shown in Fig. 6.41. For level flight, the

load factor is zero, and it increases parabolically with the increase of bank angle, in

Fig. 6.42(a). Power is slightly overestimated for roll angles φ below about 40◦ .

For higher values of φ, the predictions still follow the general trend of the flight

test data, but because of the much higher slope of the curve, the overpredictions are

higher. The right turns tend to require slightly higher power than the corresponding

left turns.

The main rotor collective, δcol , shows an excellent correlation with the flight

Figure 6.43(a) shows the lateral cyclic, δlat as a function of roll angle φ. While

for straight and level flight (φ = 0◦ ) δlat is slightly underpredicted, the agreement

The longitudinal cyclic, δlon , shown in Fig. 6.43(b), is not predicted very ac-

curately, and the variations with φ show a different trend from the flight test data.

Figure 6.44(a) shows the pitch attitude, θ. Although the trend in the flight test

data is somewhat captured in the left turns, the predictions are poor. Comparing

192

this figure with Fig. 6.43(b), it can be seen that wherever the longitudinal cyclic is

underpredicted, producing a more forward tilt of the tip path plane, the body shows

a larger nose-up pitch attitude θ. This suggests that the modeling of the fuselage

results, might be a primary reason for the discrepancy with the flight test data.

The pitch rate q is shown in Fig. 6.44(b). The theoretical results are in excel-

lent agreement with the flight test data, for the entire range of turn rates. No flight

test data was available to correlate the roll rate predictions, shown in Fig. 6.44(c).

The predicted results indicate that a negative mild roll rate needs to be applied for

most turn cases. Right turns, in particular, need a more negative roll rate than the

left turns.

Figure 6.45(a) presents the pedal setting, δped The predictions are reasonably

accurate. Note that for the higher turn rates, where the overall power required

predicted by the model is much higher than the flight test data, the tail rotor

predictions decrease well below the expected experimental values to counteract the

in the tail rotor collective, with 0% corresponding to 2.69 in to the left, or 29.9

degrees of tail rotor collective, and 100% corresponding to 2.69 in to the right, or

Finally, Fig. 6.45(b) shows the yaw rate r as a function of the roll angle φ.

The yaw rate is an almost linear function of the roll angle, and is in an excellent

Figures 6.46, 6.47 and 6.48 show, respectively, the side, rear and top views

193

of the tip vortex geometry for turn rates of ψ̇ = 0, ±5, ±15 and ±25 deg/sec.

At ψ̇ = 5 deg/sec (right turn), the angular rates of the body, in roll, pitch and

yaw, are still very small, with values of p = −0.2 deg/sec (Fig. 6.44(c)), q = 1.3

deg/sec (Fig. 6.44(b)) and r = 4.8 deg/sec (Fig. 6.45(b)). Such small angular rates

do not deform the wake geometry significantly (the only appreciable difference is

due to the orientation of the rotor). Similarly, at ψ̇ = −5 deg/sec (left turn), the

angular rates are small and the wake geometry resembles that in level flight. As the

turn rate increases to ψ̇ = ±15 deg/sec, the angular rates increase slightly, however

not enough to produce major deformations on the geometry described by the tip

vortices. The roll rate p is around 1.0 deg/sec or less, too small to have any effect

on the wake geometry. The rear view, Fig. 6.47, shows a very small difference in the

geometry due to the effect of the pitch and yaw rates (as the effect of the roll rate

is negligible), particularly in the right turn, which shows that the far wake remains

slightly closer to the plane of the rotor than it does in level flight. The pitch rate

q is slightly larger, around 9.0 deg/sec for both left and right turns, and the side

view, Fig. 6.46, shows that the vortex filaments are compressed closer together in

the upper part of the wake (closer to the rotor), while the distance between them

increases in the lower part of the wake. The yaw rate r is about ±11.5 deg/sec for

right and left turns respectively and the top view of the wake, Fig. 6.48, shows that

the edges of the wake (as seen from the top) are more compressed that in level flight.

For turns at ψ̇ = ±25 deg/sec, the angular rates are large enough to significantly

deform the wake geometry. The pitch rate, q, takes values around 20 deg/sec for

left and right turns and its effect is to contract the wake on the top, with vortex

194

filaments much closer together than at ψ̇ = ±15 deg/sec, while further increasing

the distance between each filament at the bottom of the wake. This is seen as a

“flattening” of the free wake geometry, particularly obvious at the top and in the

advancing side as seen from the side, Fig. 6.46. The effect is similar on turns to both

sides, although the different pitch attitude gives an apparently different appearance.

The roll rate p is still small, −2.0 deg/sec for left turns and −4.2 deg/sec for right

turns. Its effect, together with the effect of the higher pitch and yaw rates, can

now be appreciated in the rear view of the vortex filaments, Fig. 6.47, which also

shows signs of the “flattening” seen from the side, as well as increased bundling

of the vortices on the advancing side for the turn to the right, and reciprocally

on the retreating side for left turns. The effect of the yaw rate, r, with values of

±14.5 deg/sec approximately, is to push the vortex filaments inwards into the wake

structure in the advancing side in the right turns. In the left turns, however, the

vortex filaments are not as changed in the advancing side as in the right turns, but

in the retreating side they are pushed closer together by effect of the yaw rate.

The induced velocities λ at turn rates of ψ̇ = 0, ±5, ±15 and ±25 deg/sec are

shown in Fig. 6.49. With increasing turn rate, the pitch rate increases nose up for

both left and right turns (see Fig. 6.44(b)). This brings the wake progressively closer

to the plane of the rotor, particularly to the rear (see Fig. 6.46), and consequently

the induced velocities increase over the entire rotor, less negative in the front and

higher and positive in the rear of the rotor, but specially in the rear where the

Figure 6.50 shows the distribution of angle of attack α over the rotor disk

195

for turn rates of 0, ±5, ±15 and ±25 deg/sec. For low turn rates, such as ψ̇ = ±5

deg/sec, the changes in the distribution of angle of attack from level flight are small

and can barely be appreciated. However, as the turn rate ψ̇ increases, the angle of

around ψ = 0◦ and in the third quarter of the rotor, between ψ = 180◦ and ψ = 270◦ ,

of attack α depends on the geometric pitch θG , given by Eq.(2.55), and the velocities

flight, the change in angle of attack is due mostly to the variation in UP caused by

the rotor with respect to the incoming airflow, which affect the distribution of UP

and UT , and the azimuthal variation of the geometric pitch of the blade (Fig. 6.64),

The lift coefficient, CL , distribution for the turns at ψ̇ = 0, ±5, ±15 and ±25

deg/sec is shown in Fig. 6.51. The lift coefficient is a function of both angle of

attack α and Mach number M (Fig. 5.1). As for the angle of attack, the difference

in the distribution in lift coefficient between level flight and turns at low rates such

the rotor increase by about 0.6, with the largest values in the same areas where

the angle of attack is maximum, i.e., around ψ = 0◦ and between ψ = 180◦ and

same turn cases at ψ̇ = 0, ±5, ±15 and ±25 deg/sec. The factor of M 2 , which

196

increases radially towards the tips and is larger on the advancing than on the re-

treating side, biases the local lift distribution towards the tip and to the advancing

side, while the areas inboard of the blade, where the Mach number is low, become

regligible. Although the lift coefficient does not significantly change between level

flight and ψ̇ = ±5 deg/sec, a slightly larger Mach number at this turn rates trans-

lates into some minor but appreciable differences in the elemental lift values near the

tips. As the turn rate increases, so do the values of CL M 2 near the tips, particularly

around ψ = 0◦ and between ψ = 180◦ and ψ = 270◦ where CL is larger, but also

the advancing side where the effect of the maximum Mach number is dominant, and

The local flap moment distribution, rCL M 2 , which indicates the contribution

of the local lift to the flapping angle, is shown in Fig. 6.53 for the turns considered

at ψ̇ = 0, ±5, ±15 and ±25 deg/sec. The values of rCL M 2 mimic those of the

elemental lift CL M 2 in Fig. 6.52, although slightly more biased towards the tip by

0, ±5, ±15 and ±25 deg/sec. At low turn rates, the drag coefficient over the rotor

is very similar to the values at level flight. At ψ̇ = ±25 deg/sec, however, the high

values of angle of attack α (Fig. 6.50) and Mach number that are found towards the

tip of the blade increase the drag coefficient from an average of 0.01 in level flight

to 0.03 and larger near the tips. Around ψ = 90◦ , particularly, the drag coefficient

197

reaches value of 0.08.

±25 deg/sec, is shown in Fig. 6.55. As for the elemental lift, the effect of the square

of the Mach number is to bias the values towards those areas where M is highest,

i.e., the tips and mostly in the advancing side. The values of CD M 2 mirror those of

CD with the biased described above, reaching values at ψ̇ = ±25 deg/sec of 0.035

The elemental induced torque, rCL M 2 sin φ, seen in Fig. 6.56 for turns at

ψ̇ = 0, ±5, ±15 and ±25 deg/sec, increases in the rear of the rotor with increasing

turn rate. At low turn rates of ψ̇ = ±5 deg/sec, the change from level flight is

at the rear of the rotor the values reach 0.04. The first part of the induced torque,

rCL M 2 , seen in Fig. 6.53, is higher towards the tip, with values particularly high in

the advancing side. However, the second term, sin φ, depends directly on the inflow,

shown in Fig. 6.49. The induced velocities in the front of the rotor are particularly

low, close to zero, negating the effect of the high rCL M 2 near the tip at ψ = 90◦ .

In the rear, however, it reaches values around 0.2, with the consequent effect of

Figure 6.57 shows the elemental profile torque, rCD M 2 cos φ, at turn rates of

±5, ±15 and ±25 deg/sec. While the profile torque remains low for the low and

moderate turn rates, at ±25 deg/sec the profile drag increases considerably near the

tip in the advancing side (see Fig. 6.55) increasing the values of the profile torque

near the tips, particularly around ψ = 90◦ where it reaches a value of 0.03.

198

For moderate turns, the primary change from level flight is in the induced

power which is a function of the increased rotor thrust. The increase is higher in the

right turns (as it is for in the induced velocities as well), and thus the power required

is higher in right turns than in left turns (Fig. 6.42(a)). However, the higher induced

torque is not the only cause for the increase in power required. Like in the induced

torque, the profile torque is also higher in the right turns, contributing further to

Figure 6.58 shows the pitching moment coefficient, CM over the rotor disk for

turns at ψ̇ = 0, ±5, ±15 and ±25 deg/sec. The moment coefficient is a function of

the angle of attack α and the Mach number M , as described by Fig. 5.3. For low

positive values of α, the moment coefficient is almost constant and negative. For

low Mach numbers, beyond α = 10◦ , CM increases into the positive realm before

dropping with large nose-down values. For higher Mach numbers, the same behavior

is observed, but it occurs for angles of attack higher than α = 5◦ . For very high

Mach numbers approaching unity, the moment coefficient quickly drops into large

negative values even for very low angles of attack. At low turn rates, the moment

coefficient looks very similar to level flight, i.e., a moderate negative value around

−0.01 for much of the rotor disk, except for very negative values close to the tip

on the advancing side, where the angle of attack is negative and the Mach number

is large, and two regions of higher, close to positive values around ψ = 0◦ and

on the outer part of the third quadrant, between ψ = 180◦ and ψ = 270◦ , where

the angle of attack is higher. As the turn rate increases, both left and right, so

does the angle of attack, and therefore the moment coefficient becomes positive in

199

the areas described above. Moreover, the angle of attack also increases over the

advancing side, where the Mach number is also higher, and the moment coefficient

increases overall, particularly in a pocket around ψ = 90◦ . For very high turn rates,

ψ̇ = ±25◦ , much of the outer part of the rotor is operating at angles of attack and

the angle of attack at the tip around ψ = 90◦ reaches values around 6◦ to 8◦ (higher

on the right turns) at a point where the Mach number reaches M = 0.7. This brings

the moment coefficient beyond the maximum positive value into the large negative

drop, and therefore in this area CM takes very low values around −0.03 (over a

The elemental pitching moment, CM M 2 , is shown in Fig. 6.59 for the same

cases at ψ̇ = 0, ±5, ±15 and ±25 deg/sec. The factor M 2 makes only the values

near the tips important, as inboard of the blade the Mach number tends to zero.

The local moment, therefore, is almost uniform for the inner part of the rotor for

almost all the turn cases, with negative values close to zero. Like CM , the values

of CM M 2 increase into the positive range in the areas where the angle of attack is

highest, on the third quadrant and at the rear of the rotor. And as for CM , a pocket

of very large negative local moment is found around ψ = 90◦ where both α and M

are high.

200

6.3.2 Blade dynamics

Figure 6.60 shows the variation of the equivalent flapping angle β at the tip

of the blade as a function of the azimuth angle ψ for turns at ψ̇ = 0, ±5, ±15 and

±25 deg/sec. As the turn rate increases, both left and right, the equivalent flap

angle β increases. For low turn rates, such as ψ̇ = ±5 deg/sec, the flap angle is

only slightly larger than in level flight between ψ = 270◦ and ψ = 90◦ , i.e., on the

rear half of the rotor. For turn rates of ψ̇ = ±15 deg/sec, the flap angle is about

1 degree larger in the front half of the rotor, while in the rear the increase reaches

about 2 degrees over the level flight flap angle. For much higher turn rates, ψ̇ = ±25

deg/sec, the same is observed, with an additional 2 degrees more flap angle that at

ψ̇ = ±15 deg/sec turns. Moreover, for all turn rates, the flap angle in right turns is

slightly higher than for the left counterpart. The magnitude and phase of the flap

time histories are shown in Fig. 6.61. The first harmonic is much larger than the

higher harmonics, with values increases from 0.8 degrees in level flight to about 1.7

The local flap moment shown in Fig. 6.53 contributes to understanding the

flapping response in turning flight. As the turn rate increases, the overall value

around the azimuth. At the higher rates, it is possible to appreciate that the flap

moment is slightly higher on the right turns, and thus the flap angle is also slightly

higher. At the lower rates, the flap moment is very similar on both left and right

201

turns, although slightly higher around ψ = 0◦ on the right turns, producing slightly

larger flap angle in the rear of the rotor. In general, the lower values of rCL M 2 on

the advancing side, particularly on the second quadrant of the rotor, decrease the

flap angle; the higher values on the third quadrant increase the flap on the blade,

which increases between ψ = 180◦ and ψ = 270◦ ; on the third quadrant the local

flap decreases significantly, and thus the flap goes down, before increasing again at

the rear of the rotor due to the higher values of rCL M 2 around ψ = 0◦ .

Figure 6.62 shows the the equivalent lag angle ζ at the tip as a function of

azimuth angle ψ for the turn rates shown above. In level flight, the variation of

the lag angle over the azimuth is almost flat, with very small amplitude, and a

mean value about ζ = −3.5◦ . For moderate turns at ψ̇ = ±5 deg/sec, the lag time

history is very similar, although the lag is slightly larger (more negative) by about

0.5◦ . At ψ̇ = ±15 deg/sec, the lag angle increases over the entire azimuth, to a

mean value around ζ = −4.5◦ , with a slightly larger amplitude of about 0.7◦ . At

about ζ = −8.5◦ to −9◦ . The amplitude of the response also increases to about 1.5◦ .

In all cases, the lag angle is higher on the right turns than on the left ones, with

the difference increasing with ψ̇. The response at all turn rates is mostly a 1/rev

variation, with the magnitude of the first harmonic much larger than the higher

The local drag distribution, shown in Fig. 6.55, explains the lag response obtain

at different turn rates. In level flight and at low turn rates, the drag is very low and

almost uniform (with values around 0.005), except for slightly larger values on the

202

advancing side. The lag, therefore, is low and has almost the same value around

the azimuth, with slightly larger values on the advancing side where the drag is

higher. As the turn rate increases to ψ̇ = ±15 deg/sec, a pocket of higher drag

on the advancing side (close to 0.01) and overall slightly larger values around the

azimuth increase both the average lag displacement around the azimuth as well as

the difference between the values on the advancing and retreating side. At ψ̇ = ±25

deg/sec, the drag increases considerably, mostly on the advancing side and around

ψ = 0◦ , specially at around ψ = 90◦ , where the local drag reaches values of 0.035.

advancing side where the drag is higher. At this turn rate, the difference between

left and right turns is visible, with both larger values of CD M 2 over slightly larger

Figure 6.64 shows the elastic torsional deflection φ at the tip of the blade as

a function of azimuth angle. The magnitude and phase of the first three torsional

harmonics are shown in Fig. 6.65. The elastic torsional response shows multiple local

maxima and minima, particularly at the higher turn rates. In level flight and low

turn rates of ψ̇ = ±5 deg/sec, φ is maximum a the rear of the rotor, then decreases

up to ψ = 100◦ , and then increases slowly until it reaches the maximum at the rear

of the rotor, with a small local peak around ψ = 270◦ . At these turn rates, the

first harmonic, much larger than the higher harmonics in magnitude, dominates the

of ψ̇ = ±15 deg/sec, the magnitude of the first harmonics does not change, although

the proportion between the first and the second and third harmonics decreases. The

203

response is similar to the one at lower rates in magnitude, although several local

At much higher turn rates, ψ̇ = ±25 deg/sec, the magnitude of the first harmonic

still remains the same, but the second and third harmonic increase in magnitude

to about 2/3 of that of the first harmonic. The overall amplitude and mean value

of the elastic torsion response increase, and the oscillations at the 2/rev and 3/rev

The moment coefficient distribution at different turn rates, shown in Fig. 6.58,

deg/sec, there are two regions of mildly positive CM , around ψ = 0 and on the third

quadrant of the rotor, between ψ = 180◦ and ψ = 270◦ , that cause the blade to

twist up. At higher turn rates, ψ̇ = ±15 deg/sec, the pitching moment in these two

areas increases, with another area of positive CM appearing between ψ = 90◦ and

ψ = 120◦ ; these three areas produce local maxima of the torsional response, which

decreases in between. At very high turn rates, ψ̇ = ±25 deg/sec, there are regions

of both large positive and negative (ψ ≈ 90◦ ), which excite the higher harmonic

204

6.4 Climbing and descending flight

The results for the climb and descent cases presented in this section were obtained

at a speed V = 60 kts. Results were calculated for a range of flight path angles,

Figure 6.66(a) compares the predicted power required by the main rotor with the

experimental values of Ref. 10. Only five experimental values are available. The

point at Vc = 500 ft/min is well captured. For the three points at descent rates of

Vc = 1000 ft/min and above, the predictions are in good agreement with the test

data, taking into account the scatter in the measurements. For this range of Vc the

power required by the rotor goes to zero, which can be interpreted as an autorotative

Figure 6.66(b) shows the collective setting δcol for the same flight conditions

as in Fig. 6.66(a). The level of agreement with flight test data is mostly the same,

Figures 6.67(a) and 6.67(b) show the longitudinal and lateral stick positions

δlon and δlat . The longitudinal cyclic is in reasonable agreement with the experimen-

tal data. In climbing flight, δlon remains fairly flat; in descents it decreases slightly.

The lateral cyclic δlat is also fairly constant with flight path angle, both in climbs

The helicopter pitch attitude θ is shown in Fig. 6.68(a). The simulation over-

205

predicts θ by 1 to 3 degrees, although the predicted behavior has the same trend

as the flight test data. Two factors need to be taken into account. First, extrap-

olating from Fig. 6.4, the level flight value at 60 kts appears to be overpredicted

by nearly 2 degrees, and this offset probably affects the cases obtained at different

climb and descents angles. Moreover, the longitudinal cyclic δlon (Fig. 6.67(a)) is

underpredicted for the entire range of flight path angles, with higher deviation from

the experimental values in climb than in descent, where the difference in the pitch

up pitch of the fuselage with additional δlon to produce a forward tilt of the TPP (see

Fig. 6.98). A possible explanation could be the lack of modeling of the rotor-fuselage

Finally, Fig. 6.68(b) shows the delta position, δped . As the power decreases

with increasing descent angle, the torque that the tail has to counteract is reduced,

and therefore less tail rotor collective is needed. The reverse occurs in climbing

The behavior of the wake in climbing flight is shown in Figs. 6.69 through 6.71.

Figure 6.69 shows the side view of the wake geometry for climb angles from γ = 0◦

to 18◦ . For low climb angles, the wake remains close to the rotor. As γ increases, the

vortex filaments are convected farther down from the rotor, and their influence at

the rotor plane is reduced. This is also clear in the rear view of the wake geometry,

Fig. 6.70.The top view, Fig. 6.71, shows that with increasing γ the vortex filaments

grow apart and the wake stretches farther downstream. Recall that all the wakes

206

are calculated with the same vortex length of 1440 degrees.

The side view of the wake geometry for descent angles from γ = 0◦ to −18◦

is shown in Fig. 6.72. For level flight, the wake is trailed behind and below the

rotor, but as the flight path angle decreases beyond approximately γ = −9◦ the

vortex filaments start to get closer to the rotor and eventually cross above the rotor

plane. For steeper descents, γ < −18◦ , the entire wake is above the plane of the

rotor, organized again into a cleaner helical structure and the helicopter is in the

windmill brake state. Figure 6.73 shows the rear views of the wake. In level flight,

there is some roll up of the wake, mostly on the advancing side. As the angle of

descent increases, the roll up becomes more pronounced, with the near rolled up

bundles initially over the rotor plane and then progressively moving below the rotor

plane. At high descent angles the wake is fully over the rotor plane and the roll

up becomes less evident, eventually The top views are shown in Fig. 6.74. For low

values of γ, the vortex filaments remain bundled and tight, but as the descent angle

increases and the wake goes over the rotor plane, the distance between the vortex

filaments increases and the interaction between filaments is reduced until it almost

disappears.

The inflow distribution for the different climb cases is shown in Fig. 6.75. As

the climb angle increases and the wake is washed farther down from the rotor plane,

the influence of the vortex filaments decreases and so does their effect on the inflow

at the rear of the rotor. With increasing climb, the wake skew angle is also reduced,

the vortices become closer at the front of the rotor and the induced velocities become

207

the inflow barely changes in the rear of the rotor, while it becomes less negative in

the front. Between γ = 9◦ and γ = 12◦ , the increased proximity of the vortices to

the front of the rotor increases the inflow there, while decreasing it in the rear of

the rotor.

The inflow distribution in descending flight is shown in Fig. 6.76. For low

angles of descent, the vortex filaments get closer to the rotor plane, and their effect

becomes noticeable in the induced velocities. For flight path angles between γ = −3◦

and γ = −6◦ , some ripples can be seen in the inflow distribution over the rotor disk.

These are areas where blade-vortex interactions (BVI) are likely to occur. As the

descent angle becomes steeper the upwash induced by the forward velocity decreases,

while the effect of the wake vortices now crossing over the rotor plane is reflected

in higher but irregular induced velocities in the rear half of the rotor. Eventually,

these irregular patterns seen in the inflow disappear when the helicopter enters the

windmill brake state and the wake reforms into a clean helical structure over the

rotor.

The changes in the distribution of the induced velocities with climb angle

are reflected in the angle of attack, and therefore in the lift distribution over the

rotor. However, notice that, for climbing and descending flight, not only the inflow,

but also the climb or descent velocities play an important role in determining the

local blade angle of attack α, as well as the changes in the forward velocity due

to the increasing flight path angle and any changes in the geometric pitch of the

blade. Figure 6.77 shows the distribution of angle of attack α over the rotor for

different climb angles. At low values of γ, the angle of attack α decreases slightly

208

almost uniformly over the rotor, i.e., the distribution looks very similar to the level

flight angle of attack, only with slightly lower values of α, with lower values α

on the advancing side than the retreating side, and two areas of large values of

angle of attack continues to decrease on the advancing side and through the third

and ψ = 330◦ , while the area of higher α around ψ = 0◦ remains the same. As γ

increases to 12◦ and beyond, the angle of attack remains constant on the advancing

side, with values around 1◦ ; on the retreating side, the value α decreases slightly

in the areas where it was highest at ψ = 0◦ and between ψ = 180◦ and ψ = 270◦ ,

while it increases to the same value between ψ = 270◦ and ψ = 360◦ , so that the

as the flight path angle increases, so does the vertical component of the velocity of

the helicopter. The incoming airflow, opposite in direction and therefore downward,

increases the total value of UP at each blade section almost uniformly over the rotor.

At the same time, keeping the total speed constant while increasing the flight path

angle decreases the forward speed by a factor of cos γ. On the advancing side, the

reduced forward speed decreases the tangential component of the velocity seen at

the blade section, UT , while on the advancing side UT increases with lower forward

the ratio of UP to UT decreases the value of the angle of attack α, and vice-versa.

On the advancing side, UP increases and UT decreases, therefore the ratio of the two

209

increases, and the angle of attack decreases slightly. On the retreating side, however,

ratio decreases, increasing the angle of attack, particularly around ψ = 270◦ , where

the maximum UT occurs. The distribution of angle of attack α over the rotor in

descending flight is shown in Fig. 6.78. With decreasing flight path angle, the angle

of attack does not change significantly at the tips of the rotor. Inboard of the

blade, the angle of attack increases, mostly on the retreating side, and becomes

of attack occur near the root of the blade, specially on the retreating side around

ψ = 270◦ , which decrease radially outward of the blade. In descents, the change

that occurs in UP is the opposite as in climbs, the incoming upward flow decreases

the value of UP , while the changes in the tangential component UT mirror those

during climbs, i.e., on the advancing side, the reduced forward speed decreases the

tangential component of the velocity seen at the blade section, UT , and the opposite

happens on the retreating side. The closer to the root, the lower the value of UT (as

the tips, however, the changes in the ratio of UP to UT are small and the angle of

Figure 6.79 shows the lift coefficient distribution for climbing angles between

attack, α, and the Mach number, M (Fig. 5.1). As explained in previous sections,

The lift coefficient is highest in those areas where the angle of attack is higher,

biased towards the tips where the Mach number, which increases the CL for a given

210

α, is highest. As with the angle of attack, CL has to areas of high values around

ψ = 0◦ and in the third quadrant of the rotor in level flight. As the flight path

angle increases, the lower values of angle of attack decrease the CL , particularly on

the advancing side, and at higher values of the climb angle beyond γ = 6◦ , the lift

increases on the retreating side. The lift coefficient for descending cases between

the lift coefficient increases inboard of the blade, like the angle of attack, reaching

The factor of M 2 biases the values of the local lift towards the tip of the blades,

the local lift increases on the retreating side and around ψ = 0◦ , where α is highest,

and around ψ = 90◦ , where the Mach number is maximum. For steeper climbs,

however, as the forward speed decreases by a factor of cos γ, the Mach number

In descents, Fig. 6.82, the areas where the lift coefficient is heavily biased towards

the root of the blades, are now very small because of the factor of M 2 , which is

approaches zero towards the root. At the same time, M 2 brings forward the areas

of higher local lift. In particular, at low rates of descents, two areas of high CL M 2

can be identified in the first and third quadrant of the rotor, between ψ = 0◦ and

ψ = 90◦ and between ψ = 180◦ and ψ = 270◦ respectively. As the descent angle

increases, particularly beyond γ = −9◦ , the local lift on the first quadrant remains

211

Figure 6.83 shows the contribution of the local flap moment, rCL M 2 , for the

same climb angles used above. The factor of the nondimensional radius biases the

values towards the tip even more. The highest values of rCL M 2 are found in the

front-retreating side and around ψ = 0◦ , in the outer quarter of the blade. In the

descent cases, Fig. 6.84, the same is observed, i.e., the areas where CL M 2 is higher

are highlighted and biased the closer to the tips by the effect of the product of r.

Figure 6.85 shows the drag coefficient, CD for different climb angles between

and Mach number M (Fig. 5.2). As described earlier, for angles of attack lower than

α = 5◦ and Mach numbers up to M = 0.6, the drag coefficient has an almost constant

value around CD = 0.008. At low climb angles, the drag coefficient distribution is

almost constant at this value, except for the regions in the third quadrant of the

rotor, between ψ = 180◦ and ψ = 270◦ , and around ψ = 0◦ , where α reaches values

reaching values of around CD = 0.03. In this area, the Mach number seen locally

by the blade is between 0.6 and 0.7. The angle of attack, which is very low for the

entire advancing side with values between α = 0◦ and 1◦ near the tips, becomes

much larger for a small pocket around ψ = 90◦ , where it reaches values about

α = 5◦ . At this high Mach numbers and angles of attack, a small change in angle of

attack increases the drag considerably (Fig. 5.2). For higher climb angles, the ares of

higher angle of attack barely reach 5◦ , and occur where the Mach number is low, and

therefore the drag coefficient remains constant for the entire rotor at CD = 0.008.

The drag coefficient distribution in descents is shown in Fig. 6.86. Throughout the

212

descent cases, the regions of high angle of attack occur near the root, where the

Mach number is low, and therefore the drag coefficient is only dependent on α. As

shown in Fig. 6.78, the angle of attack increases near the root with increasing descent

angle. Similarly, the drag coefficient CD mirrors the areas of high α, becoming very

Figure 6.87 shows the local drag coefficient, CD M 2 , in climbing flight. The

effect of M 2 is to bias the local drag towards the tip, particularly on the advancing

side, where the Mach number is highest, making the areas inboard of the blade,

where M is low, negligible. The areas where the drag coefficient in climb is highest

occur where the Mach number is low, making the local drag CD M 2 negligible there.

In descents, shown in Fig. 6.88, the same is observed: the high values of drag

coefficient near the root become almost negligible when multiplied by M 2 , and the

maximum local drag is found near the tips on the advancing side, where the Mach

Figure 6.89 shows the elemental induced torque, rCL M 2 sin φ, for climb angles

between γ = 0◦ and γ = 18◦ . The first part of the expression rCL M 2 sin φ has

already been explained in Fig. 6.83. The second part, the sine of the induced angle

φ = tan−1 ( UUPT ), depends on several factors. On one side, the climb velocity increases

the perpendicular velocity UP over the whole rotor, while the climb angle also affects

the tangential component UT , which is slightly larger on the retreating side than

(Fig. 6.75) also affect the perpendicular velocity UP . Therefore, the most significant

variation in φ over the rotor is due to the changes in the inflow. The inflow is roughly

213

close to zero or negative in the front of the rotor disk, and positive and increasing

towards the rear of the rotor. As a consequence, the induced torque is mostly zero

(or slightly negative) in the front of the rotor, despite rCL M 2 being large near the

tips in the front of the rotor. The values of the induced torque increase towards the

tips in the rear of the rotor, although biased towards the retreating side, reaching

a maximum value of 0.025. In terms of climb angle, the induced velocities do not

change a lot in climbs up to γ = 9◦ , and beyond this climb angle the inflow becomes

slightly less negative in the front of the rotor, and lower in the rear of the rotor.

increases with increasing flight path angle up to γ = 9◦ , and the values decrease

slightly for higher climb angles. The induced torque in descending flight is shown in

Fig. 6.90. The induced velocities are much lower in descending flight than in climbs

(Fig. 6.76), and thus the induced torque is also lower. As in climb, the inflow is

close to zero or negative in the front of the rotor, and therefore the induced torque

is also minimum there, and it increases towards the tips at the rear of the rotor.

Figure 6.91 shows the elemental profile torque, rCD M 2 cos φ, in climbing flight.

Since φ is a small angle, the main factor affecting the profile torque is the local drag.

The distribution of profile torque follows that of the local drag shown in Fig. 6.87,

although the factor of the local radius r biases the profile torque more towards the

tips, while it becomes even more negligible closer to the root. In low climb angles, up

to γ = 9◦ , the profile torque shows two distinctive areas where higher values occur,

on the advancing side near the tips where the Mach number is maximum, and on

214

the third quadrant, between ψ = 280◦ and ψ = 270◦ , also near the tip, where the

drag is high due to the higher angle of attack. Win increasing climb angle, the angle

of attack decreases, and so does the drag, therefore the local profile torque on the

third quadrant diminishes. On the advancing side near the tips, the profile torque

remains high, although the lower Mach number found as the climb angle increases

reduced the maximum value of profile torque reached in that area. In descents, the

profile torque Fig. 6.92 behaves the same way. As the local drag, shown in Fig. 6.88,

the maximum profile torque is found where the Mach number is highest, near the

The increase in power in climbing flight seen in Fig. 6.66(a) is clearly not

associated with profile torque. The maximum profile torque found on a small band

near the tip on the advancing side is half the magnitude of the maximum induced

torque, which occurs over a much larger area of the rotor. Therefore, the higher

induced torque seen as the climb angle increases is the cause for the higher power

required in climb, and the decrease in induce torque has γ decreases therefore reduces

the need for power in descending flight. Observe that the induced torque is larger

for the case of γ = 9◦ than it is at γ = 12◦ , which explains the higher power required

to operate at that particular flight condition that it does at slightly higher climb

angles.

Figure 6.93 shows the moment coefficient CM over the rotor for different climb

angles. In level flight, there are two areas where the moment coefficient is larger,

i.e., around ψ = 0◦ in the outboard half of the blade, and in the third quadrant

of the rotor, between ψ = 180◦ and ψ = 270◦ , also in the outer half of the blade.

215

At γ = 3◦ the distribution looks very similar to that in level flight. At γ = 6◦

areas described above, while it remains similar in the rest of the rotor. For higher

climb angles, the moment coefficient decreases in those localized areas, although

it increases slightly in the rest of the retreating side. In all cases, at the very tip

of the blade where the angle of attack is zero or negative, the moment coefficient

becomes much more negative. Figure 5.3 is useful to understand the changes in

CM with increasing flight path angle. As stated before, the moment coefficient

is a function of both the angle of attack α and the Mach number M . For very

high Mach numbers, M ≈ 0.7 and above, the moment coefficient CM is negative

for the entire range of positive angle of attack values, particularly low for α > 5◦ .

negative value for angles of attack up to 10◦ , then it increases till becoming slightly

positive before plunging into large nose-down moments. For the Mach numbers

the increase and drop in CM occurs at increasingly lower angles of attack, reaching

also increasingly more positive values before dropping into very negative values.

Figure 6.94 illustrates how the different values of angle of attack and Mach number

affect the moment coefficient seen in Fig. 6.93. It shows the moment coefficient for

function of the angle of attack, over-imposed over a section of Fig. 5.3, for two of

the climb cases, γ = 6◦ and γ = 12◦ . On the advancing side (ψ = 92◦ in the figure.),

the Mach numbers achieved are large at these outboard radial stations, however the

216

angle of attack is low in both climb cases, and therefore the moment coefficient is

negative and moderate. The moment coefficient increases around the azimuth for

both climb angles and radial stations, reaching a maximum at around ψ = 233◦ . In

both climb cases, the Mach number is higher for the outer blade station, although

the angle of attack is lower there, and therefore the moment coefficient is higher

at the r = 0.85R station than at r = 0.90R. Except for the outer blade station

of attack is more than 1◦ larger than at γ = 12◦ , and this difference increases the

moment coefficient considerably. The moment then goes down at all radial stations

and climb cases for the rest of the azimuth locations, although it increases again at

the rear of the rotor, at ψ = 358◦ . The increase, however, is not enough to make

the moment coefficient positive in all cases but one, in the r = 0.85R radial location

As in climb, the moment coefficient depends mostly on the angle of attack for except

for at large values of α and M . As the descent angle increases, the angle of attack

grows close to the root, and the moment coefficient is large, even positive, in that

area. On the advancing side, the angle of attack is lower, but the Mach number is

higher, and therefore an area of larger CM can be found in the first quarter of the

rotor.

Figure 6.96 shows the elemental moment coefficient CM M 2 for the same climb

cases as before. The product of the square of the Mach number biases the values

of the local moment towards the tip, and more on the advancing side than the

retreating side. The local moment is almost constant at a moderate negative value

217

around CM ≈ −0.001 for all the inner half of the rotor. On the outer half on the

advancing side, some low negative values are reached, due to the negative angle of

attack, while on the advancing side some areas of higher CM are observed, specially

on the third quadrant where the angle of attack is higher. In descents, shown in

Fig. 6.97, the high values of CM inboard of the blade are negated by the low Mach

number inboard of the blade, and, similarly to the climb cases, CM M 2 is mostly

uniform in the inner half of the rotor. In the outer half, slightly higher values are

reached on the third quadrant for the lower descent angles, and both on the first

The equivalent flapping angle β at the blade tip as a function of azimuth is shown in

Fig. 6.98, for several climb and descent angles, including level flight for reference. As

the flight path angle increases, the phase of the flapping response changes, but the

increase, so does the magnitude of the flapping response, which for all the cases

Figure 6.99 shows magnitude and phase of the first three harmonics of the

flapping response. As can be seen from Fig. 6.98, the magnitudes of the harmonics

are much larger in the descent than in the climb cases. Throughout the flight path

range, the 1/rev response is dominant. In climbs, however, the ratio between the

218

1/rev and the 2/rev flapping is larger (particularly in the γ = 6◦ and γ = 12◦

climbs), and the presence of the 2/rev excitation is visible in the shape of the flap

curves.

A possible explanation for the shape of the flap distribution at the different

flight path angles can be obtained from Figs. 6.83 and 6.84, which shows the dis-

tribution of flapping moment, rCL M 2 , over the rotor disk for different values of

γ. Areas of high rCL M 2 contribute to increase the flap displacement, while low

rCL M 2 values decrease it. In level flight, the flap moment starts to decrease from

equivalent flap angle at the tip, therefore, decreases on the advancing side. Around

ψ = 150◦ , the flapping moment starts to increase again, reaching a maximum around

ψ = 270◦ , then decreases considerably for almost the entire fourth quadrant of the

rotor, before increasing again around ψ = 0◦ . The flap angle, therefore, increases

on the third quadrant of the rotor, and then decreases towards the rear of the rotor.

Because of this multiple variation around the azimuth, the flap higher harmonics

are being excited. In climbing flight, the local flapping moment has has a similar

distribution than in level flight, although the areas of high rCL M 2 around ψ = 0◦

and on the third quadrant are slightly higher in magnitude, affecting the flap history.

In descents, the local maxima of rCL M 2 do not reach as high values as in climbing

flight. However, an area of negative contribution to the flap angle occurs on the

second quadrant of the rotor, between ψ = 90◦ and ψ = 180◦ . This large flap-down

moment decreases the lag angle, which reaches a minimum during descents around

ψ = 180◦ , while the high flap moment on the third quadrant increases again. The

219

negative dip increases with descent angle, which increases the amplitude of the flap

time history. Although the mean flap value in climbs is higher, due to the higher

average value of the local flap moment, the amplitude in descents is much higher

caused by the larger variation in the values of rCL M 2 around the azimuth.

Figure 6.100 shows the equivalent lag angle ζ at the tip as a function of azimuth

angle ψ for the climb and descent cases being considered. All climb and descent cases

look very similar in shape, although the mean value of each time history changes

considerably with γ. The amplitude of the sinusoidal variation of the equivalent tip

lag is also very similar in all cases. Increasing the climb angle, however, increases

the mean value of the lag, which doubles by the time γ = 18◦ . Increasing the

climb angle also changes the phase of the distribution slightly, with the maximum

flight path angle decreases, the average lag angle also decreases, and the point of

maximum lag occurs later over the azimuth. Figure 6.101 shows the magnitude and

phase of the first three harmonics of the equivalent lag angle. In climbing flight,

the magnitude of all three harmonics is pretty similar, specially for the higher climb

angles. However, these are much lower than the magnitude of the constant lag

coefficient (ζ0 = −6.6852, ζ1c = −0.083403, ζ1s = −0.11343, ζ2c = −0.14286, etc.,

at γ = 18◦ ), and therefore the lag time histories appear almost flat. In descents,

the magnitude of the first harmonic is clearly dominant over the higher hamonics,

and although smaller than the constant coefficient as in climb, the lag time histories

220

elemental drag, shown in Figs. 6.87 and 6.88. The local drag in the outer quarter of

the blade is much larger in climbs than in descents, which explains the much larger

the outer part of the rotor with climb angle, mostly on the retreating side, which

increases the lag displacement for the steeper climbs. The variation over the rotor

is very moderate, and thus the lag displacement does not change considerably as

a function of azimuth. On the advancing side, the higher drag found at the tip

increases the lag, which increases again on the retreating side where the drag is

lower. In descents, the difference between the advancing and retreating side is more

defined, with higher drag on the advancing side increasing the lag displacement and

lower drag on the retreating side that decreases the lag. The maximum drag on the

advancing side decreases as the descent angle increases, and therefore the average

lag decreases too. While in descents the variation in drag is clearly a 1/rev variation,

perturbing mostly the first harmonic, in climbs the drag distribution exciting the

Figures 6.102 and 6.103 show, respectively, the elastic torsional deflection φ

at the blade tip as a function of azimuth angle, and the magnitude and phase of

the first three torsional harmonics.The magnitude of the first harmonic increases

progressively from almost 0.4 degrees at γ = 18◦ to slightly more than 1 degree

at γ = −18◦ . Its phase is almost constant with γ. The magnitude of the second

harmonic also increases as γ goes from 18◦ to −18◦ , and remains below 0.2 degrees

except at γ = 6◦ . Its phase has more noticeable variations, which determine the

221

The distribution of pitching moment over the rotor, shown in Figs. 6.93 and 6.95

for γ = 0◦ , ±6◦ , ±12◦ and ±18◦ , helps understand the behavior of the torsion of the

blade. In descents, the very negative pitching moment found on the advancing side

excites the first harmonic, which is much larger than the higher harmonics; the

elastic torsion response shows a large nose-down deflection on the advancing side

corresponding to this negative pitching moment. In climbs, the advancing side does

not show such large negative values, and the regions of positive CM , which occur

at ψ = 0◦ and on the third quadrant of the rotor, excite the higher harmonics of

This section contains the results of simulations performed at high rates of descent

and low speeds, for velocities V = 20, 25 and 30 kts and descent angles between γ =

−21◦ and γ = −50◦ . Several simulations were carried out progressively approaching

the onset of VRS, with a combination of velocities and flight path angles. True

cannot be properly studied as a trim condition. However, the onset and incipient

stages of VRS can be explored as trim states, both to understand the underlying

physical mechanism, and to study the numerical behavior of the trim algorithm.

Figure 6.104 shows the VRS boundary as obtained from the experiments of

Drees and Hendal (Ref. 1). The boundary is delimited by a combination of forward

222

speed ratios Vx /Vh and descent speeds Vz /Vh , where Vx , Vz and Vh are, respectively,

the velocity in the plain of the rotor, the velocity perpendicular to the plane of the

rotor and a velocity scale, namely the ideal inflow in hover from momentum theory

p

T /2ρA. Within this boundary, signs of VRS were observed in the experiments.

Superimposed are the cases analyzed in the present study, which approach or cross

over the boundary that determines the onset of VRS. While the experimental data

are not rigorously applicable for the full-scale helicopter configuration used in the

present study, Fig. 6.104 is still used here to provide a general indication of the stage

of each descent case. The different markers in the figure denote the flight conditions

analyzed. In general, the slower and steeper the trajectory, the deeper the rotor

VRS. Figure 6.105 shows the variation of main rotor power required and collective

for flight path angles between γ = 0◦ and γ = −50◦ at V = 20 kts.The steeper the

the need for more iterations of the trim procedure, and a higher residual in the trim

obtained a steady-state solution with the current model. The results show that the

power required decreases almost linearly with γ (much steeper descents, outside the

range of the present study, would be necessary for conditions of ”power settling” to

Figure 6.106 shows the fuselage angle of attack, fuselage pitch angle and the

223

kts. The aerodynamic angle of attack of the fuselage αF increases linearly with the

descent angle, while the pitch angle θ is almost constant with γ and increases by less

than two degrees as the flight path angle goes from γ = 0◦ to γ = −50◦ . The rotor

responds by tilting forward, with a positive longitudinal flapping β1c (Fig. 6.108(a))

that increases by about two degrees as the flight path angle decreases to γ = −50◦ .

Figure 6.107 shows the fuselage roll angle, lateral cyclic and pedal setting and

Fig. 6.108(b) shows the lateral flapping, for γ = 0◦ to γ = −50◦ at V = 20 kts. Roll

angle φ and lateral flapping β1s are not significantly affected. The conditions in the

present study are not severe enough to affect the lateral motion of the helicopter,

although deep in the VRS both pitch and roll oscillations are usually observed.

The pedal position increases by 10% (right) with γ to maintain heading due to the

change in torque.

Figure 6.109 shows side and top views of the wake geometry for γ = −30◦ ,

−40◦ and −50◦ at V = 20 kts. The incipient stages of VRS are characterized by

an initial convection of the vortex filaments over the rotor tip path plane, which

are subsequently pulled back under the rotor, and some instability develops in the

form of vortex ”bundling” (Ref. 83) of the filaments downstream of the rotor. The

bundling is clearly visible for all the values of γ in this figure. As the descents

become steeper, the formation of these vortex rings occurs closer to the rotor and

they are not convected as far from it. If the rate of descent were to increase any

further, the accumulation of vorticity would reach the rotor plane and the rotor

224

Figure 6.110 shows the distribution of both the induced velocities λ and the

perpendicular component of the sectional velocity UP over the rotor for the same

descent angles at V = 20 kts. For the lower values of γ, the upward flow starts

convecting the vortex filaments over the rotor, but they stay close to the tip path

plane and their proximity can be identified as regions of drastic changes in the

induced velocities. With increasing descent rate, the vortices move closer to the

rotor plane and the induced velocities increase towards the tip. In the meantime, the

upward flow due to the high rates of descents significantly decrease the perpendicular

velocity at the blade section, which is composed of the inflow, the velocities due to

translation and rotation and the those due to blade flexibility. Figure 6.111 shows

the angle of attack α and the lift coefficient CL corresponding to these cases. In

the stage where the vortex filaments are first convected above the rotor, a region of

very high angle of attack appears, and stall occurs at that location. The increase

in upward flow that reduces UP increases the overall angle of attack, except for the

region on the advancing side at γ = −30◦ where the tip vortex is crossing the rotor.

The increase is slightly more pronounced on the retreating side, as the tangential

path angle, than on the advancing side. The lift coefficient distribution, depicts the

occurrence of stall inboard of the blade and some regions of negative lift close to the

Figure 6.112 shows the elemental lift and induced torque distributions for the

above flight conditions. The figure also shows the distribution of induced torque,

which is observed to decrease the steeper the descent. The lower UP that occurs

225

with the lower values of γ decreases the value of the induced angle φ, decreasing the

induced torque. As a consequence, the power required by the main rotor, shown

in Fig. 6.105, decreases as well in these steep descents. In addition, the tangential

component UT increases on the retreating side, and decreases on the advancing side.

Therefore, the decrease in φ is more pronounced on the retreating side than on the

Figure 6.113 shows the local flap moment and the pitching moment coefficients

at γ = −30◦ , −40◦ and −50◦ . The flapping moment rCL M 2 is similar to the

those values close to the root negligible, and those with a larger moment arm towards

2

the tip are the only ones contributing to the flap motion. As for CM , the flap moment

is higher towards the tips in the rear of the rotor disk, between ψ = 270◦ and ψ = 90◦

increases. However, the difference between the values in the rear and the rear of the

rotor disk increases slightly with descent angle. The moment coefficient CM , which

is a function of angle of attack and Mach number, shows particularly high, even

positive, values towards the tip, specially on the retreating side, but also around

ψ = 90◦ . The Mach number, at this high descent angles and low speed, is almost

axi-symmetric, with peak values around M = 0.6 at the tips. For such value of M ,

angles of attack higher than 4◦ produce positive CM (Fig. 5.3, while for the slightly

−40◦ and −50◦ and V = 20 kts is shown in Fig. 6.114. All three cases have the

226

same phase and similar average flap value as well, although the amplitude of the

response increases with the steeper descent angles. The local flap moment, shown

in Fig. 6.113, shows that the lift contributes to an increased flap in the rear of the

rotor, while in the front the contribution is minimum reducing the flap angle. The

higher difference between the local flap moment in the front and the rear of the

rotor seen at the steeper descents causes the larger magnitude in the flap response.

The elastic torsional displacement for the same steep descent cases is shown

in Fig. 6.115. Unlike the flapping response, the elastic twist decreases in amplitude

with the higher descent angles. The response in all three cases looks similar, only

slight off phase, with multiple harmonics being excited. The variation in the moment

This section presents results for steep descending turns, with a speed V = 40 kts

along the trajectory, flight path angle γ = 40◦ and with turn rates from ψ̇ = −40

Figure 6.116(a) as a function of turn rate for turns from ψ̇ = −40 deg/sec to

with turn rate until it reaches values around 1.5 at ψ̇ = ±40 deg/sec, moderately

larger on the left turns than on the right ones. The load factor increases slower in

descending turns that in level turns due to the lower thrust required.

227

Figure 6.116(b) shows the main rotor power required as a function of turn

rate. The power required increases with turn rate, both for left and right turns.

For straight flight at γ = 40◦ the power required is almost zero and the helicopter

more power and thrust are required to overcome the centrifugal force and perform

the turns. For turn rates of up to ψ̇ = 30 deg/sec, the increment in the power

required is similar for both left and right turns. At rates of ψ̇ = ±40 deg/sec the

left turn requires about 100 hp more than the right turn.

Figure 6.116(c) shows the main rotor collective, δcol , for the different turn rates.

Unlike the power, the right turns show a higher requirement for collective than the

left turns, as much as 5 % at the higher turn rates. The change in magnitude of

the required collective with varying turn rate is similar to that of the turns at level

flight (Fig. 6.42(b)), only about 20 % lower due to the descending flight condition

The lateral cyclic, δlat , helicopter roll angle, φ, and roll rate, p, as a function

of turn rate are shown in Fig. 6.117. The lateral cyclic, which in level fight turns

decreased for turns to both left and right directions (Fig. 6.43(a)), presents a larger

magnitude decrease in the port diving turns, although in the starboard dives it

increases (although note that besides the flight path angle, the diving turns are also

performed at lower speed, 40 kts vs. 60 kts for the level turns). The helicopter

roll angle increases proportionally to the turn rate, both with left and right turns.

While the helicopter is highly banked in the more severe cases, the angle does not

change at a high rate. Up to turn rates of ψ̇ = ±20 deg/sec, the roll rate is around

228

0. On the right turns, it increases to around p = 8 deg/sec at ψ̇ = 40 deg/sec, while

on the left turns it reaches values slightly higher in magnitude, but negative, around

The longitudinal cyclic, δlon , helicopter pitch angle, θ, and pitch rate, q, are

shown in Fig. 6.118. The longitudinal cyclic increases initially with the lower rates

of turn, both left and right, which follows the trends of the experimental level flight

turns (Fig. 6.43(b)), but at the highest turn rates computed, ψ̇ = ±40 deg/sec, the

it needs to continuously increase its nose-down pitch angle to maintain the flight

path angle. Up to turn rates of ψ̇ = ±30 deg/sec, the pitch attitude is very similar

in both the left and right turns. However, at ψ̇ = ±40 deg/sec the pitch attitude

of the left turn is much lower than the right equivalent one. The pitch rate needed

to turn is more than double the value in magnitude than the roll rate, increasing to

Figure 6.119 shows the pedal setting, δped , and yaw rate, r, required for the

different turn rates. Although no fight test data is available for this flight condition,

a comparison with the level fight turn predictions for power and pedal (Figs. 6.42(a)

and 6.45(a)) seems to indicate that also in the diving turns the power at the higher

turn rates might be slightly overpredicted, and that the pedal setting for these

extreme diving turns might be lower than the trends observed at more moderate turn

rates to compensate for the additional torque (like mentioned before, a percentage

decrease in the pedal setting indicates an increase in the pedal to the left, which

229

increases the tail rotor collective). The yaw rate, which controls the actual turn of

the helicopter in the helical path of the coordinated turn, is the largest of the three

Figures 6.120 through 6.125 show the resulting geometry side, rear and top

views for the descending turns being studied at V = 40 kts, γ = −40◦ and ψ̇ =

0, ±10, ±20, ±30 and ±40 deg/sec. These turns take place at a very steep angle

of descent, therefore the wake vortices are convected above the rotor disk, even as

the rate of turn, and therefore rotor thrust, increases. Vortex bundling begins to

appear for turn rates higher than about ψ̇ = 30 deg/sec. With increasing turn rate,

both right and left, the tip path plane pitches down, bringing the wake vortices

closer to the rear of the rotor. The main differences between left and right turns are

noticed at the higher turn rates. Right turns vortex bundling is more evident on

the advancing side, but almost disappears on the retreating side, while in the left

turn the vortex bundle is more evident, and extends further, with a higher area of

Figure 6.126 shows the induced velocities for turns at ψ̇ = 0 deg/sec, ±20

deg/sec and ±40 deg/sec. The induced velocities in the left turns are clearly higher

than in the right turn at the same rate. While the inflow very near tip of the

front and retreating side is slightly higher in the right turn, the left turn shows

Figure 6.127 shows the angle of attack for the same turns at ψ̇ = 0 deg/sec,

±20 deg/sec and ±40 deg/sec. At these high rates of descent, the upward flow

230

considerably. However, the high angular rates experienced at the higher turn rates

also affect the perpendicular and tangential velocities UP and UT , and therefore the

induced angle φ and the angle of attack, to a great extent. As a result, the angle of

attack changes from an almost axi-symmetric (slightly higher on the retreating side)

on the second quadrant of the rotor on the right turns, and on the first quadrant on

the left turns. The maximum angle of attack reached is higher on the right turns,

where it reaches 15◦ in a region close to the root between ψ = 150◦ and ψ = 180◦ ,

while on the left turns the maximum values reached inboard of the blade are about

11◦ .

shown in Fig. 6.128. In the regions where the angle of attack is highest, the Mach

number is low, therefore the lift coefficient depends mostly on α. As the angle of

attack, the lift coefficient is maximum close to the root and decreases radially for

the low turn rates, although at high turn rates the higher values are found on the

second quadrant of the rotor disk for the right turns and on the first quadrant for

The elemental lift distribution, CL M 2 , is given in Fig. 6.129. The values close

to the root become less important by the factor of M 2 , which goes to zero at the

root. Only the values in regions of high dynamic pressure, close to the tips and

specially on the advancing side, are important. At low turn rates, the areas where

the lift coefficient is slightly higher begin to be noticeable, and occur on the second

quadrant for the right turn, where values of CL M 2 = 0.17 are found, and on the first

231

quadrant for the left turns, where similar values are reached, although over a slightly

elemental lift almost double, reaching CL M 2 = 0.3, again in the same areas. At

these rates, however, the region where CL M 2 is highest is larger on the left turns,

although over the entire rotor, the values are higher on the right turns.

deg/sec and ±40 deg/sec. At low turn rates, the combinations of angles of attack

and Mach numbers experienced are low enough to produce an almost constant local

drag. At the highest turn rates, however, the drag increases considerably, due to the

On the right turns, the elemental drag reaches CD M 2 = 0.012 over the outer half of

the second quadrant of the rotor, while on the left turns, similar values are reached

and ±40 deg/sec, is shown in Fig. 6.131. While the right turn has higher induced

torque requirements in the third quadrant of the rotor, between ψ = 180◦ and

ψ = 270◦ approximately, the left turn has overall higher values, which translate into

the higher power requirements in the left turn. The value of the induced torque

depends both on the local lift at each point on the rotor and the induced angle, i.e.,

φ = UP /UT . The induced angle, which changes considerably between the left and

right turns due to the different angular rates, drives the main differences between

the two cases. On the right turns, the induced angle is negative on the advancing

side, which coupled with the large values of CL M 2 found on the second quadrant,

232

produces very low negative values there. On the retreating side, the elemental lift

was lower, but φ is larger, producing large positive values of the induced torque.

On the left turns, CL M 2 is not as large on the second quadrant of the rotor disk,

therefore while negative, the induced torque is not as low. The highest value of

the induced torque occurs on the rear of the rotor, increasing the induced torque,

Figure 6.132 shows the elemental profile torque, rCD M 2 cos φ at ψ̇ = 0 deg/sec,

±20 deg/sec and ±40 deg/sec. The profile torque depends on the cosine of the

induced angle, which is close to one, therefore the distribution resembles that of the

elemental drag, with the torque increasing considerably at the higher turn rates due

to areas of high drag on the second quadrant for the right turns and on the first

The slightly higher power required for the left turns is justified by the slightly

higher values of the induced torque, as the overall profile torque is very similar

(maximum value is slightly higher on the left turn, although the right turn has

The results in the present study were obtained using a 2.5 GHz Quad G5 (two dual

core PowerPC processors) Macintosh computer with 1GB of RAM. The code is not

currently designed to exploit parallelism and therefore the resultant effect from a

233

Trim convergence is sensitive to the initial solution for both the model with

dynamic inflow and that with the free wake. Even with the former, for which

convergence is relatively easier, starting from a poor initial solution often leads to

as possible.

The trim procedure is also sensitive to flight condition. At high speeds, i.e. for

µ > 0.18, it is relatively easy to achieve convergence. Hover and low speed condi-

tions are usually more difficult and computationally expensive. The recommended

procedure is to compute trim at moderately high speeds (e.g. 0.23 < µ < 0.28)

reduction of 10 kts at each step usually proved successful in this study. At each

speed, the initial solution is the converged solution for the previous speed.

The overall CPU time required to achieve convergence was primarily driven

by two factors, namely, the increased computational effort required to obtain a trim

solution in hover and low speed, and the CPU time required to achieve convergence

of the free wake geometry for each evaluation of the trim equations. Recall, however,

that the function evaluations required to build the Jacobian matrix did not include

a recalculation of the wake, as explain in Section 4.5.5). Convergence for the free

wake model is determined by observing the L2 norm of the induced velocities over

the rotor. The root mean square (RM S) change in the inflow is calculated using,

v

u Nψ Nr

1 u X X

RM S = t (λ(ψj , rk )n − λ(ψj , rk )n−1 )2 (6.1)

Nψ Nr ψ:j=1 r:k=1

where Nψ is the number of blade azimuthal steps in one revolution and Nr is the

234

number of radial points used along the blade.

At lower speeds, the vortex filaments remain closer to the rotor and their

influence is larger, whereas at higher speeds, the vortices are quickly washed away

behind the rotor, and their influence is lessened. As a consequence, the inflow norm

The figures that follow are representative of the computation time and the

resentative of high speed flight, obtaining a trim solution from an initial solution

which 92 are required to build the Jacobian twice using one-sided finite differences.

During those 92 evaluations, the free wake is not evaluated, but on each of the

remaining 23, the free wake calculations are carried out until convergence. The

complete trim calculation for the coupled rotor-fuselage-wake system requires ap-

proximately about 1.5 hours. At V = 60 kts, the process requires a similar effort,

i.e., 120 evaluations of the trim equations of motion, of which 92 are needed to com-

cost increases considerably, as it the solver needs to compute a new Jacobian matrix

6 times, requiring 306 evaluations of the trim equations of motion. However, hover

is the most difficult case, as it requires 316 evaluations of the equations of motion

and 6 builds of the Jacobian to obtain a solution. In terms of time, the process

the turn takes place, and the same arguments apply. In addition, the higher the

235

turn rate, the more difficult achieving convergence becomes. As for the low speed

cases, the approach to converge results at high turn rates consists in progressively

incrementing the turn rate and using each solution as the starting estimate for the

Similarly, climbing and descending flight cases are susceptible to the speed

at which the flight condition occurs, and for a given speed, the higher the flight

path angle, the more difficult convergence becomes. Incremental progress is also

used to achieve convergence at high flight path angles, either in climb or descent.

might be necessary.

Finally, a note on the accuracy of the converged solution. The algebraic equa-

tion solver tries to reduce the residual of the equations of motion by reducing the

difference between the approximate present solution X and a previous solution Xsol

The tolerance used in the present study is 10−6 . By meeting this criterion, the norm

of the residuals of the equations of motion usually becomes very small. However, as

the speed is decreased, not only it takes longer to obtain a trim solution, but the

solver returns a solution with larger and larger residuals. The procedure used in the

present study was the following: if a solution was returned with a overall residual

norm greater than 10−3 , the accuracy was considered insufficient and therefore the

solution was discarded and the trim solution process repeated. At high speeds,

236

this was not an issue, and he solutions obtained resulted in residual norms of the

order of 10−8 . At hover and low speeds, however, poor convergence was frequently

a problem, but solutions were found with residual norms of the order of 10−4 , which

cedure is that speed and extent of convergence are very sensitive to the value of

the vortex core growth parameter δ Eq.(3.18). The smaller the vortex core growth

value, the smaller the core size and the higher the velocity field that each filament

sees because of the other filaments. This increases the strength of their mutual

interactions, and this results in a more difficult convergence of the free wake, and

therefore of the entire trim procedure. Increasing that value would improve conver-

gence, as has been found many times in the literature (Refs. 48, 52). However, in

this study the vortex core growth was not used as a tuning parameter. A realistic

value determined from experimental results (Ref. 119) was used instead, even if it

Another important observation regarding the cost of computation for the trim

calculation is that the present approach, described in Section 4.5.5, eliminates the

calls to the free wake during the calculation of the Jacobian matrix. In early stages

of the present study, as well as in the study that serves as a baseline for the present

work (Ref. 6), the wake was used to calculate the inflow at each evaluation of the

trim equations, whether these were to compute the Jacobian matrix or to advance

the solution (see Appendix C for a schematic of the trim procedure in Ref. 6). Con-

sidering that a single Jacobian (in the present configuration, as its size depends on

237

the number of modes and other factors) requires 46 evaluations of the trim equa-

tions, converging the wake for each of those evaluations constituted a significantly

large cost of computation. However, it was discovered that the free wake model

is quite insensitive to the small perturbations used to obtain the derivatives that

form the Jacobian (1% of the quantity being perturbed). Therefore, the solution

attempted consisted on calculating the free wake inflow only on the iteration that

establishes the baseline for the Jacobian calculations and on each trim calculation

used to obtain a solution, but not in those evaluations of the trim equations required

to build the Jacobian, for which the inflow from the last solution computed is used.

The present study also simplifies the convergence of the inflow-circulation problem.

Previous stages of this work and the earlier related studies (Ref. 6) used an assumed

value of the inflow to calculate the bound vortex circulation (see Appendix C). The

inflow was converged with this fixed value of Γ. But the inflow and circulation

are mutually dependent, therefore the new inflow had to be used to re-evaluate the

circulation, and this double-loop was repeated until the inflow converged. In the

method for the calculation of the inflow has permitted that the inflow calculation

be replaced by a single loop, which updates Γ for each new value of the inflow,

therefore reducing the number of calls to the free wake dramatically. In terms of

cost, this translates into a reduction of an order of magnitude, since the previous

approach needed 24 to 30 hours for a simple solution (in the above mentioned ma-

chine, much more in earlier Powermacs with G4 PowerPC processors), while now it

238

3500

Main Rotor Power Required (HP)

Simulation

2500

2000

1500

1000

500

0

0 50 100 150 200

Velocity, V (kts)

Figure 6.1: Main rotor power required, QM R , as a function of speed, for a straight

blade configuration.

239

120

Main Rotor Collective Stick, δcol (%)

100

80

60

40

20 Flight test

Simulation

0

0 50 100 150 200

Velocity, V (kts)

Figure 6.2: Main rotor collective required as a function of speed, for a straight blade

configuration.

240

100

Flight test

Longitudinal Stick Position, δlon (%)

Simulation

80

60

Aft

Forward

40

20

0

0 50 100 150 200

Velocity, V (kts)

Figure 6.3: Main rotor longitudinal stick position as a function of speed, for a

241

6

2

Pitch Attitude, θ (deg)

-2

-4

-6 Flight test

Simulation

-8

0 50 100 150 200

Velocity, V (kts)

Figure 6.4: Helicopter pitch attitude as a function of speed, for a straight blade

configuration.

242

100

Flight test

Simulation

80

Lateral Stick Position, δlat (%)

Right

60

40

Left

20

0

0 50 100 150 200

Velocity, V (kts)

Figure 6.5: Main rotor lateral stick position as a function of speed, for a straight

blade configuration.

243

2

Flight test

Simulation

1

Roll Angle, φ (deg)

-1

-2

-3

0 50 100 150 200

Velocity, V (kts)

Figure 6.6: Helicopter bank angle as a function of speed, for a straight blade con-

figuration.

244

100

Flight test

Simulation

80

Pedal Position, δped (%)

Right

60

40

Left

20

0

0 50 100 150 200

Velocity, V (kts)

Figure 6.7: Pedal position as a function of speed, for a straight blade configuration.

245

1.5 1.0

1.0

0.5

0.5

z/R

z/R

0.0 0.0

−0.5

−0.5

−1.0

−1.5 −1.0

−1.5 −1.0 −0.5 0.0 0.5 1.0 1.5 −1 0 1 2 3

x/R x/R

1.0

0.5

z/R

0.0

−0.5

−1.0

−1 0 1 2 3

x/R

Figure 6.8: Side view of the wake geometry at V = 1, 40 and 100 kts.

246

1.5 1.0

1.0

0.5

0.5

z/R

z/R

0.0 0.0

−0.5

−0.5

−1.0

−1.5 −1.0

−1.5 −1.0 −0.5 0.0 0.5 1.0 1.5 −1.5 −1.0 −0.5 0.0 0.5 1.0 1.5

y/R y/R

1.0

0.5

z/R

0.0

−0.5

−1.0

−1.5 −1.0 −0.5 0.0 0.5 1.0 1.5

y/R

Figure 6.9: Rear view of the wake geometry at V = 1, 40 and 100 kts.

247

1.5 1.5

1.0 1.0

0.5 0.5

y/R

y/R

0.0 0.0

−0.5 −0.5

−1.0 −1.0

−1.5 −1.5

−1.5 −1.0 −0.5 0.0 0.5 1.0 1.5 −1 0 1 2 3

x/R x/R

1.5

1.0

0.5

y/R

0.0

−0.5

−1.0

−1.5

−1 0 1 2 3

x/R

Figure 6.10: Top view of the wake geometry at V = 1, 40 and 100 kts.

248

0.15 0.15

90 90

1 1

120 60 120 60

0.8 0.8

150 30 150 30

0.4 0.4

180 0 180 0

0 0

−0.05 −0.05

270 270

−0.1 −0.1

0.15

90

1

120 60

0.8

0.6 0.1

150 30

0.4

0.2 0.05

180 0

210 330

−0.05

240 300

270

−0.1

249

15 15

90 90

1 1

120 60 120 60

0.8 0.8

0.6 0.6

150 30 10 150 30 10

0.4 0.4

0.2 0.2

180 0 5180 0 5

270 270

−5 −5

(a) Angle of attack, α (deg), V = 1 kt. (b) Angle of attack, α (deg), V = 40 kts.

15

90

1

120 60

0.8

0.6

150 30 10

0.4

0.2

180 0 5

210 330 0

240 300

270

−5

Figure 6.12: Distribution of aerodynamic angle of attack over the rotor disk.

250

90 90

1 1

120 60 1 120 60 1

0.8 0.8

0.8 0.8

0.6 0.6

150 30 0.6 150 30 0.6

0.4 0.4

0.4 0.4

0.2 0.2

0.2 0.2

180 0 180 0

0 0

−0.2 −0.2

−0.6 −0.6

270 270

−1 −1

90

1

120 60 1

0.8

0.8

0.6

150 30 0.6

0.4

0.4

0.2

0.2

180 0

0

−0.2

−0.6

270

−1

251

0.25 0.25

90 90

1 1

120 60 0.2 120 60 0.2

0.8 0.8

150 30 150 30

0.4 0.1 0.4 0.1

180 0 0 180 0 0

−0.05 −0.05

−0.1 −0.1

210 330 210 330

−0.15 −0.15

270 270

−0.25 −0.25

0.25

90

1

120 60 0.2

0.8

0.6 0.15

150 30

0.4 0.1

0.2 0.05

180 0 0

−0.05

−0.1

210 330

−0.15

270

−0.25

252

90 90

1 0.2 1 0.2

120 60 120 60

0.8 0.8

0.15 0.15

0.6 0.6

150 30 150 30

0.4 0.4

0.1 0.1

0.2 0.2

0.05 0.05

180 0 180 0

0 0

−0.1 −0.1

240 300 240 300

270 270

90

1 0.2

120 60

0.8

0.15

0.6

150 30

0.4

0.1

0.2

0.05

180 0

−0.1

240 300

270

Figure 6.15: Distribution of local flap moment, rCL M 2 , over the rotor disk.

253

90 90

1 1

120 60 120 60

0.8 0.8

0.025 0.025

0.6 0.6

150 30 150 30

0.4 0.4

0.02 0.02

0.2 0.2

180 0 180

0.015 0 0.015

0.01 0.01

210 330 210 330

0.005 0.005

240 300 240 300

270 270

90

1

120 60

0.8

0.025

0.6

150 30

0.4

0.02

0.2

180 0 0.015

0.01

210 330

0.005

240 300

270

254

0.02 0.02

90 90

1 1

120 60 0.018 120 60 0.018

0.8 0.8

150 30 150 30

0.4 0.014 0.4 0.014

0.008 0.008

0.006 0.006

210 330 210 330

0.004 0.004

270 270

0 0

0.02

90

1

120 60 0.018

0.8

0.6 0.016

150 30

0.4 0.014

0.2 0.012

180 0 0.01

0.008

0.006

210 330

0.004

270

0

255

0.02 0.02

90 90

1 1

120 60 120 60

0.8 0.015 0.8 0.015

0.6 0.6

150 30 0.01 150 30 0.01

0.4 0.4

180 0 0 180 0 0

−0.005 −0.005

−0.015 −0.015

240 300 240 300

270 270

−0.02 −0.02

(a) rCL M 2 sin φ (N-D), V = 1 kt. (b) rCL M 2 sin φ (N-D), V = 40 kts.

0.02

90

1

120 60

0.8 0.015

0.6

150 30 0.01

0.4

0.2 0.005

180 0 0

−0.005

−0.015

240 300

270

−0.02

Figure 6.18: Elemental induced torque distribution, rCL M 2 sin φ, over the rotor

disk.

256

0.02 0.02

90 90

1 1

120 60 0.018 120 60 0.018

0.8 0.8

150 30 150 30

0.4 0.014 0.4 0.014

0.008 0.008

0.006 0.006

210 330 210 330

0.004 0.004

270 270

0 0

(a) rCD M 2 cos φ (N-D), V = 1 kt. (b) rCD M 2 cos φ (N-D), V = 40 kts.

0.02

90

1

120 60 0.018

0.8

0.6 0.016

150 30

0.4 0.014

0.2 0.012

180 0 0.01

0.008

0.006

210 330

0.004

270

0

Figure 6.19: Elemental profile torque distribution, rCD M 2 cos φ, over the rotor disk.

257

0.01 0.01

90 90

1 1

120 60 120 60

0.8 0.005 0.8 0.005

0.6 0.6

150 30 0 150 30 0

0.4 0.4

−0.015 −0.015

−0.025 −0.025

240 300 240 300

270 270

−0.03 −0.03

0.01

90

1

120 60

0.8 0.005

0.6

150 30 0

0.4

0.2 −0.005

180 0 −0.01

−0.015

−0.025

240 300

270

−0.03

Figure 6.20: Distribution of pitching moment coefficient, CM , over the rotor disk.

258

−3 −3

x 10 x 10

5 5

90 90

1 1

120 60 120 60

0.8 0.8

0.6 0.6

150 30 0 150 30 0

0.4 0.4

0.2 0.2

180 0 −5 180 0 −5

270 270

−15 −15

−3

x 10

5

90

1

120 60

0.8

0.6

150 30 0

0.4

0.2

180 0 −5

240 300

270

−15

Figure 6.21: Distribution of elemental pitching moment, CM M 2 , over the rotor disk.

259

6

1 kt

40 kt

5 100 kt

Equivalent flap angle, β (deg)

1

0 90 180 270 360

Azimuth angle, ψ (deg)

260

2.0

1st harmonic

2nd harmonic

3rd harmonic

1.5

1.0

0.5

0.0

1 40 100

Speed, V (kts)

(a) Magnitude.

180

1st harmonic

2nd harmonic

Phase of flapping harmonics (deg)

3rd harmonic

90

-90

-180

1 40 100

Speed, V (kts)

(b) Phase.

Figure 6.23: Magnitude and phase of the first three flapping harmonics at V = 1

261

-3.0

-3.5

Equivalent lag angle, ζ (deg)

-4.0

-4.5

-5.0

-5.5

-6.0

1 kt

-6.5 40 kt

100 kt

-7.0

0 90 180 270 360

Azimuth angle, ψ (deg)

262

0.20

1st harmonic

2nd harmonic

3rd harmonic

0.15

0.10

0.05

0.00

1 40 100

Speed, V (kts)

(a) Magnitude.

180

Phase of lag harmonics (deg)

90

-90

1st harmonic

2nd harmonic

3rd harmonic

-180

1 40 100

Speed, V (kts)

(b) Phase.

Figure 6.25: Magnitude and phase of the first three lag harmonics at V = 1 kt, 40

263

1.5

1 kt

Elastic torsion at the blade tip, φ (deg)

40 kt

1.0 100 kt

0.5

0.0

-0.5

-1.0

-1.5

0 90 180 270 360

Azimuth angle, ψ (deg)

Figure 6.26: Blade tip elastic torsion as a function of azimuth at V = 1 kt, 40 kts

264

1.0

1st harmonic

2nd harmonic

3rd harmonic

0.8

0.6

0.4

0.2

0.0

1 40 100

Speed, V (kts)

(a) Magnitude.

180

Phase of torsion harmonics (deg)

90

-90

1st harmonic

2nd harmonic

3rd harmonic

-180

1 40 100

Speed, V (kts)

(b) Phase.

Figure 6.27: Magnitude and phase of the first three torsion harmonics at V = 1 kt,

265

2500

Main Rotor Power Required (HP)

2000

1500

1000

500

Flight test

Rigid

Flexible

0

0 20 40 60 80 100 120 140

Velocity (kts)

Figure 6.28: Main rotor power required, QM R , as a function of speed, with a rigid

266

100

Main Rotor Collective Position (%)

80

60

40

20

Flight test

2 modes

5 modes

0

0 20 40 60 80 100 120 140

Velocity (kt)

Figure 6.29: Main rotor collective as a function of speed, with a rigid blade model

267

0.15

90

1

120 60

0.8

0.6 0.1

150 30

0.4

0.2 0.05

180 0

210 330

−0.05

240 300

270

−0.1

−3

x 10

90 14

1

120 60

0.8

12

0.6

150 30

0.4 10

0.2

8

180 0

4

210 330

2

240 300

270 0

Figure 6.30: Comparison of the induced inflow distribution with the rigid and flexible

268

6

5

Equivalent flap angle, β (deg)

2

Flexible

Rigid

1

0 90 180 270 360

Azimuth angle, ψ (deg)

269

-0.02

180 deg (front) 0 deg (rear)

-0.04

VX (Non-Dimensional)

-0.06

-0.08

-0.10

-0.12

Flexible

Rigid

-0.14

-1 -0.5 0 0.5 1

x/R (Non-Dimensional)

(a) Vx

0.020

180 deg (front) 0 deg (rear)

0.015

0.010

VZ (Non-Dimensional)

0.005

0.000

-0.005

-0.010

-0.015 Flexible

Rigid

-0.020

-1 -0.5 0 0.5 1

x/R (Non-Dimensional)

(b) Vz

Figure 6.32: Comparison of the x and z-components of the velocity seen by the

blade section for a longitudinal cross section of the rotor, with the rigid and flexible

270

50

Flexible

40 Rigid

30

Circulation, Γ

20

10

-10

180 deg (front) 0 deg (rear)

-20

-1 -0.5 0 0.5 1

x/R (Non-Dimensional)

Figure 6.33: Comparison of the bound circulation for a longitudinal cross section of

the rotor, with the rigid and flexible blade models; V = 1 kt.

271

3500

Flight test

3000 Straight tip

Swept tip

Main rotor power required (HP)

2500

2000

1500

1000

500

0

0 50 100 150 200

Velocity, V (kts)

Figure 6.34: Main rotor power required, QM R , as a function of speed with a time-

marching free wake model with a straight and swept tip blade models.

272

120

Flight test

Straight tip

100

Swept tip

Main rotor collective (%)

80

60

40

20

0

0 50 100 150 200

Velocity, V (kts)

Figure 6.35: Main rotor collective as a function of speed with a straight and swept

273

90 90

1 1

120 60 120 60

0.8 0.02 0.8 0.02

0.6 0.6

150 30 150 30

0.4 0 0.4 0

0.2 0.2

−0.02 −0.02

180 0 180 0

−0.04 −0.04

−0.06 −0.06

210 330 210 330

−0.08 −0.08

240 300 240 300

270 270

(a) rCL M 2 sin φ, V = 1 kt, straight. (b) rCL M 2 sin φ, V = 1 kt, swept.

90 90

1 1

120 60 120 60

0.8 0.02 0.8 0.02

0.6 0.6

150 30 150 30

0.4 0 0.4 0

0.2 0.2

−0.02 −0.02

180 0 180 0

−0.04 −0.04

−0.06 −0.06

210 330 210 330

−0.08 −0.08

240 300 240 300

270 270

(c) rCL M 2 sin φ, V = 60 kts, straight. (d) rCL M 2 sin φ, V = 60 kts, swept.

90 90

1 1

120 60 120 60

0.8 0.02 0.8 0.02

0.6 0.6

150 30 150 30

0.4 0 0.4 0

0.2 0.2

−0.02 −0.02

180 0 180 0

−0.04 −0.04

−0.06 −0.06

210 330 210 330

−0.08 −0.08

240 300 240 300

270 270

(e) rCL M 2 sin φ, V = 158 kts, straight. (f) rCL M 2 sin φ, V = 158 kts, swept.

Figure 6.36: Elemental induced torque, rCL M 2 sin φ, with a straight and swept tip

274

−3 −3

x 10 x 10

90 90

1 1

120 60 120 60

0.8 7 0.8 7

0.6 0.6

150 30 6 150 30 6

0.4 0.4

0.2 5 0.2 5

180 0 4 180 0 4

3 3

1 1

240 300 240 300

270 270

(a) rCD M 2 sin φ, V = 1 kt, straight. (b) rCD M 2 sin φ, V = 1 kt, swept.

−3 −3

x 10 x 10

90 90

1 1

120 60 120 60

0.8 7 0.8 7

0.6 0.6

150 30 6 150 30 6

0.4 0.4

0.2 5 0.2 5

180 0 4 180 0 4

3 3

1 1

240 300 240 300

270 270

(c) rCD M 2 sin φ, V = 60 kts, straight. (d) rCD M 2 sin φ, V = 60 kts, swept.

90 90

1 1

120 60 0.08 120 60 0.08

0.8 0.8

150 30 150 30

0.4 0.4

0.06 0.06

0.2 0.2

0.05 0.05

180 0 180 0

0.04 0.04

0.03 0.03

210 330 210 330

0.02 0.02

270 270

(e) rCD M 2 sin φ, V = 158 kts, straight. (f) rCD M 2 sin φ, V = 158 kts, swept.

Figure 6.37: Elemental profile torque, rCD M 2 cos φ, with a straight and swept tip

275

90 90

1 1

120 60 0.25 120 60 0.25

0.8 0.8

150 30 150 30

0.4 0.4

0.15 0.15

0.2 0.2

0.1 0.1

180 0 180 0

0.05 0.05

0 0

210 330 210 330

−0.05 −0.05

270 270

90 90

1 1

120 60 0.25 120 60 0.25

0.8 0.8

150 30 150 30

0.4 0.4

0.15 0.15

0.2 0.2

0.1 0.1

180 0 180 0

0.05 0.05

0 0

210 330 210 330

−0.05 −0.05

270 270

90 90

1 1

120 60 0.25 120 60 0.25

0.8 0.8

150 30 150 30

0.4 0.4

0.15 0.15

0.2 0.2

0.1 0.1

180 0 180 0

0.05 0.05

0 0

210 330 210 330

−0.05 −0.05

270 270

Figure 6.38: Induced velocities, λ, with a straight and swept tip blade models, at

276

90 90

1 1

120 60 120 60

0.8 80 0.8 80

0.6 0.6

150 30 150 30

0.4 60 0.4 60

0.2 0.2

40 40

180 0 180 0

20 20

0 0

210 330 210 330

−20 −20

240 300 240 300

270 270

90 90

1 1

120 60 120 60

0.8 80 0.8 80

0.6 0.6

150 30 150 30

0.4 60 0.4 60

0.2 0.2

40 40

180 0 180 0

20 20

0 0

210 330 210 330

−20 −20

240 300 240 300

270 270

90 90

1 1

120 60 120 60

0.8 80 0.8 80

0.6 0.6

150 30 150 30

0.4 60 0.4 60

0.2 0.2

40 40

180 0 180 0

20 20

0 0

210 330 210 330

−20 −20

240 300 240 300

270 270

Figure 6.39: Bound circulation, Γ, with a straight and swept tip blade models, at

277

60

40

20

Roll angle (deg)

-20

-40

-60

-30 -20 -10 0 10 20 30

Figure 6.40: Relation between the roll angle, φ, and the turn rate, ψ̇.

278

1.8

1.7

1.6

Load factor, nT

1.5

1.4

1.3

1.2

1.1

1.0

-60 -40 -20 0 20 40 60

Roll angle, φ (deg)

279

3000

Main Rotor Power Required (HP) Flight test

2500 Simulation

2000

1500

1000

500

0

-60 -40 -20 0 20 40 60

Roll angle, φ (deg)

100

Main Rotor Collective Stick, δcol (%)

80

60

40

20

Flight test

Simulation

0

-60 -40 -20 0 20 40 60

Roll angle, φ (deg)

Figure 6.42: Main rotor power required, QM R , and collective, δcol , as a function of

roll angle φ.

280

100

Flight test

Lateral Stick Position, δlat (%)

80 Simulation

Right

60

40

Left

20

0

-60 -40 -20 0 20 40 60

Roll angle, φ (deg)

100

Longitudinal Stick Position, δlon (%)

Flight test

80 Simulation

60

Aft

Forward

40

20

0

-60 -40 -20 0 20 40 60

Roll angle, φ (deg)

Figure 6.43: Lateral stick position, δlat , and longitudinal stick position, δlon , as a

281

10

Flight test

Simulation

5

-5

-60 -40 -20 0 20 40 60

Roll angle, φ (deg)

(a) Helicopter pitch angle, θ

30

25 Flight test

Pitch rate, q (deg/sec)

Simulation

20

15

10

-5

-60 -40 -20 0 20 40 60

Roll angle, φ (deg)

(b) Helicopter pitch rate, q

1

0

Roll rate, p (deg/sec)

-1

-2

-3

-4

Simulation

-5

-60 -40 -20 0 20 40 60

Roll angle, φ (deg)

(c) Helicopter roll rate, p

Figure 6.44: Helicopter pitch angle, θ, pitch rate, q and roll rate, p, as a function of

roll angle φ.

282

100

Flight test

Simulation

80

Pedal Position (%)

60

40

20

0

-60 -40 -20 0 20 40 60

Roll angle, φ (deg)

15

10 Flight test

Simulation

Yaw rate, r (deg/sec)

-5

-10

-15

-60 -40 -20 0 20 40 60

Roll angle, φ (deg)

Figure 6.45: Pedal position δped , and helicopter yaw rate, r, as a function of roll

angle φ.

283

0.5 0.5

0.0 0.0

z/R

z/R

−0.5 −0.5

−1.0 −1.0

(a) ψ̇ = 0 deg/sec (b) ψ̇ = 0 deg/sec

−1 0 1 2 3 −1 0 1 2 3

x/R x/R

0.5 0.5

0.0 0.0

z/R

z/R

−0.5 −0.5

−1.0 −1.0

(c) ψ̇ = −5 deg/sec (d) ψ̇ = 5 deg/sec

−1 0 1 2 3 −1 0 1 2 3

x/R x/R

0.5 0.5

0.0 0.0

z/R

z/R

−0.5 −0.5

−1.0 −1.0

(e) ψ̇ = −15 deg/sec (f) ψ̇ = 15 deg/sec

−1 0 1 2 3 −1 0 1 2 3

x/R x/R

0.5 0.5

0.0 0.0

z/R

z/R

−0.5 −0.5

−1.0 −1.0

(g) ψ̇ = −25 deg/sec (h) ψ̇ = 25 deg/sec

−1 0 1 2 3 −1 0 1 2 3

x/R x/R

Figure 6.46: Side view of the free wake tip vortex for turns at V = 60 kts.

284

0.5 0.5

0.0 0.0

z/R

z/R

−0.5 −0.5

−1.0 −1.0

(a) ψ̇ = 0 deg/sec (b) ψ̇ = 0 deg/sec

−1.0 −0.5 0.0 0.5 1.0 −1.0 −0.5 0.0 0.5 1.0

y/R y/R

0.5 0.5

0.0 0.0

z/R

z/R

−0.5 −0.5

−1.0 −1.0

(c) ψ̇ = −5 deg/sec (d) ψ̇ = 5 deg/sec

−1.0 −0.5 0.0 0.5 1.0 −1.0 −0.5 0.0 0.5 1.0

y/R y/R

0.5 0.5

0.0 0.0

z/R

z/R

−0.5 −0.5

−1.0 −1.0

(e) ψ̇ = −15 deg/sec (f) ψ̇ = 15 deg/sec

−1.0 −0.5 0.0 0.5 1.0 −1.0 −0.5 0.0 0.5 1.0

y/R y/R

0.5 0.5

0.0 0.0

z/R

z/R

−0.5 −0.5

−1.0 −1.0

(g) ψ̇ = −25 deg/sec (h) ψ̇ = 25 deg/sec

−1.0 −0.5 0.0 0.5 1.0 −1.0 −0.5 0.0 0.5 1.0

y/R y/R

Figure 6.47: Rear view of the free wake tip vortex for turns at V = 60 kts.

285

1.5 1.5

1.0 1.0

y/R

0.0 0.0

−0.5 −0.5

−1.0 −1.0

−1.5 −1.5

−1 0 1 2 3 −1 0 1 2 3

x/R x/R

1.5 1.5

1.0 1.0

0.5 0.5

y/R

y/R

0.0 0.0

−0.5 −0.5

−1.0 −1.0

−1.5 −1.5

−1 0 1 2 3 −1 0 1 2 3

x/R x/R

1.5 1.5

1.0 1.0

0.5 0.5

y/R

y/R

0.0 0.0

−0.5 −0.5

−1.0 −1.0

−1.5 −1.5

−1 0 1 2 3 −1 0 1 2 3

x/R x/R

1.5 1.5

1.0 1.0

0.5 0.5

y/R

y/R

0.0 0.0

−0.5 −0.5

−1.0 −1.0

−1.5 −1.5

−1 0 1 2 3 −1 0 1 2 3

x/R x/R

Figure 6.48: Top view of the free wake tip vortex for turns at V = 60 kts.

286

0.2 0.2

90 90

1 1

120 60 120 60

0.8 0.8

0.15 0.15

0.6 0.6

150 30 150 30

0.4 0.4

0.1 0.1

0.2 0.2

0 0

210 330 210 330

−0.05 −0.05

270 270

−0.1 −0.1

(a) ψ̇ = 0 deg/sec (b) ψ̇ = 0 deg/sec

0.2 0.2

90 90

1 1

120 60 120 60

0.8 0.8

0.15 0.15

0.6 0.6

150 30 150 30

0.4 0.4

0.1 0.1

0.2 0.2

0 0

210 330 210 330

−0.05 −0.05

270 270

−0.1 −0.1

(c) ψ̇ = −5 deg/sec (d) ψ̇ = 5 deg/sec

0.2 0.2

90 90

1 1

120 60 120 60

0.8 0.8

0.15 0.15

0.6 0.6

150 30 150 30

0.4 0.4

0.1 0.1

0.2 0.2

0 0

210 330 210 330

−0.05 −0.05

270 270

−0.1 −0.1

(e) ψ̇ = −15 deg/sec (f) ψ̇ = 15 deg/sec

0.2 0.2

90 90

1 1

120 60 120 60

0.8 0.8

0.15 0.15

0.6 0.6

150 30 150 30

0.4 0.4

0.1 0.1

0.2 0.2

0 0

210 330 210 330

−0.05 −0.05

270 270

−0.1 −0.1

(g) ψ̇ = −25 deg/sec (h) ψ̇ = 25 deg/sec

Figure 6.49: Inflow distribution for different turn rates at V = 60 kts.

287

18 18

90 90

1 1

120 60 16 120 60 16

0.8 0.8

14 14

0.6 0.6

150 30 12 150 30 12

0.4 0.4

10 10

0.2 0.2

8 8

180 0 180 0

6 6

4 4

2 2

210 330 210 330

0 0

−2 −2

240 300 240 300

−4 −4

(a) ψ̇270= 0 deg/sec (b) ψ̇270= 0 deg/sec

18 18

90 90

1 1

120 60 16 120 60 16

0.8 0.8

14 14

0.6 0.6

150 30 12 150 30 12

0.4 0.4

10 10

0.2 0.2

8 8

180 0 180 0

6 6

4 4

2 2

210 330 210 330

0 0

−2 −2

240 300 240 300

−4 −4

270 270

(c) ψ̇ = −5 deg/sec (d) ψ̇ = 5 deg/sec

18 18

90 90

1 1

120 60 16 120 60 16

0.8 0.8

14 14

0.6 0.6

150 30 12 150 30 12

0.4 0.4

10 10

0.2 0.2

8 8

180 0 180 0

6 6

4 4

2 2

210 330 210 330

0 0

−2 −2

240 300 240 300

−4 −4

(e) ψ̇ 270

= −15 deg/sec (f) ψ̇270= 15 deg/sec

18 18

90 90

1 1

120 60 16 120 60 16

0.8 0.8

14 14

0.6 0.6

150 30 12 150 30 12

0.4 0.4

10 10

0.2 0.2

8 8

180 0 180 0

6 6

4 4

2 2

210 330 210 330

0 0

−2 −2

240 300 240 300

−4 −4

(g) ψ̇ 270

= −25 deg/sec (h) ψ̇270= 25 deg/sec

Figure 6.50: Angle of attack distribution for different turn rates at V = 60 kts.

288

90 90

1 1

120 60 1 120 60 1

0.8 0.8

0.8 0.8

0.6 0.6

150 30 0.6 150 30 0.6

0.4 0.4

0.4 0.4

0.2 0.2

0.2 0.2

180 0 180 0

0 0

−0.2 −0.2

−0.6 −0.6

270 270

−1 −1

(a) ψ̇ = 0 deg/sec (b) ψ̇ = 0 deg/sec

90 90

1 1

120 60 1 120 60 1

0.8 0.8

0.8 0.8

0.6 0.6

150 30 0.6 150 30 0.6

0.4 0.4

0.4 0.4

0.2 0.2

0.2 0.2

180 0 180 0

0 0

−0.2 −0.2

−0.6 −0.6

270 270

−1 −1

(c) ψ̇ = −5 deg/sec (d) ψ̇ = 5 deg/sec

90 90

1 1

120 60 1 120 60 1

0.8 0.8

0.8 0.8

0.6 0.6

150 30 0.6 150 30 0.6

0.4 0.4

0.4 0.4

0.2 0.2

0.2 0.2

180 0 180 0

0 0

−0.2 −0.2

−0.6 −0.6

270 270

−1 −1

(e) ψ̇ = −15 deg/sec (f) ψ̇ = 15 deg/sec

90 90

1 1

120 60 1 120 60 1

0.8 0.8

0.8 0.8

0.6 0.6

150 30 0.6 150 30 0.6

0.4 0.4

0.4 0.4

0.2 0.2

0.2 0.2

180 0 180 0

0 0

−0.2 −0.2

−0.6 −0.6

270 270

−1 −1

(g) ψ̇ = −25 deg/sec (h) ψ̇ = 25 deg/sec

Figure 6.51: Lift coefficient distribution for different turn rates at V = 60 kts.

289

90 0.35 90 0.35

1 1

120 60 120 60

0.8 0.3 0.8 0.3

0.6 0.6

150 30 0.25 150 30 0.25

0.4 0.4

0.1 0.1

0 0

240 300 240 300

270 270

−0.05 −0.05

(a) ψ̇ = 0 deg/sec (b) ψ̇ = 0 deg/sec

90 0.35 90 0.35

1 1

120 60 120 60

0.8 0.3 0.8 0.3

0.6 0.6

150 30 0.25 150 30 0.25

0.4 0.4

0.1 0.1

0 0

240 300 240 300

270 270

−0.05 −0.05

(c) ψ̇ = −5 deg/sec (d) ψ̇ = 5 deg/sec

90 0.35 90 0.35

1 1

120 60 120 60

0.8 0.3 0.8 0.3

0.6 0.6

150 30 0.25 150 30 0.25

0.4 0.4

0.1 0.1

0 0

240 300 240 300

270 270

−0.05 −0.05

(e) ψ̇ = −15 deg/sec (f) ψ̇ = 15 deg/sec

90 0.35 90 0.35

1 1

120 60 120 60

0.8 0.3 0.8 0.3

0.6 0.6

150 30 0.25 150 30 0.25

0.4 0.4

0.1 0.1

0 0

240 300 240 300

270 270

−0.05 −0.05

(g) ψ̇ = −25 deg/sec (h) ψ̇ = 25 deg/sec

Figure 6.52: Elemental lift distribution for different turn rates at V = 60 kts.

290

90 90

1 1

120 60 120 60

0.8 0.3 0.8 0.3

0.6 0.6

150 30 0.25 150 30 0.25

0.4 0.4

0.1 0.1

0 0

240 300 240 300

270 270

−0.05 −0.05

(a) ψ̇ = 0 deg/sec (b) ψ̇ = 0 deg/sec

90 90

1 1

120 60 120 60

0.8 0.3 0.8 0.3

0.6 0.6

150 30 0.25 150 30 0.25

0.4 0.4

0.1 0.1

0 0

240 300 240 300

270 270

−0.05 −0.05

(c) ψ̇ = −5 deg/sec (d) ψ̇ = 5 deg/sec

90 90

1 1

120 60 120 60

0.8 0.3 0.8 0.3

0.6 0.6

150 30 0.25 150 30 0.25

0.4 0.4

0.1 0.1

0 0

240 300 240 300

270 270

−0.05 −0.05

(e) ψ̇ = −15 deg/sec (f) ψ̇ = 15 deg/sec

90 90

1 1

120 60 120 60

0.8 0.3 0.8 0.3

0.6 0.6

150 30 0.25 150 30 0.25

0.4 0.4

0.1 0.1

0 0

240 300 240 300

270 270

−0.05 −0.05

(g) ψ̇ = −25 deg/sec (h) ψ̇ = 25 deg/sec

Figure 6.53: Local flap moment distribution for different turn rates at V = 60 kts.

291

0.08 0.08

90 90

1 1

120 60 120 60

0.8 0.07 0.8 0.07

0.6 0.6

150 30 0.06 150 30 0.06

0.4 0.4

0.03 0.03

0.01 0.01

240 300 240 300

270 270

0 0

(a) ψ̇ = 0 deg/sec (b) ψ̇ = 0 deg/sec

0.08 0.08

90 90

1 1

120 60 120 60

0.8 0.07 0.8 0.07

0.6 0.6

150 30 0.06 150 30 0.06

0.4 0.4

0.03 0.03

0.01 0.01

240 300 240 300

270 270

0 0

(c) ψ̇ = −5 deg/sec (d) ψ̇ = 5 deg/sec

0.08 0.08

90 90

1 1

120 60 120 60

0.8 0.07 0.8 0.07

0.6 0.6

150 30 0.06 150 30 0.06

0.4 0.4

0.03 0.03

0.01 0.01

240 300 240 300

270 270

0 0

(e) ψ̇ = −15 deg/sec (f) ψ̇ = 15 deg/sec

0.08 0.08

90 90

1 1

120 60 120 60

0.8 0.07 0.8 0.07

0.6 0.6

150 30 0.06 150 30 0.06

0.4 0.4

0.03 0.03

0.01 0.01

240 300 240 300

270 270

0 0

(g) ψ̇ = −25 deg/sec (h) ψ̇ = 25 deg/sec

Figure 6.54: Drag coefficient distribution for different turn rates at V = 60 kts.

292

0.035 0.035

90 90

1 1

120 60 120 60

0.8 0.03 0.8 0.03

0.6 0.6

150 30 150 30

0.4 0.025 0.4 0.025

0.2 0.2

0.02 0.02

180 0 180 0

0.015 0.015

0.01 0.01

210 330 210 330

0.005 0.005

240 300 240 300

270 270

0 0

(a) ψ̇ = 0 deg/sec (b) ψ̇ = 0 deg/sec

0.035 0.035

90 90

1 1

120 60 120 60

0.8 0.03 0.8 0.03

0.6 0.6

150 30 150 30

0.4 0.025 0.4 0.025

0.2 0.2

0.02 0.02

180 0 180 0

0.015 0.015

0.01 0.01

210 330 210 330

0.005 0.005

240 300 240 300

270 270

0 0

(c) ψ̇ = −5 deg/sec (d) ψ̇ = 5 deg/sec

0.035 0.035

90 90

1 1

120 60 120 60

0.8 0.03 0.8 0.03

0.6 0.6

150 30 150 30

0.4 0.025 0.4 0.025

0.2 0.2

0.02 0.02

180 0 180 0

0.015 0.015

0.01 0.01

210 330 210 330

0.005 0.005

240 300 240 300

270 270

0 0

(e) ψ̇ = −15 deg/sec (f) ψ̇ = 15 deg/sec

0.035 0.035

90 90

1 1

120 60 120 60

0.8 0.03 0.8 0.03

0.6 0.6

150 30 150 30

0.4 0.025 0.4 0.025

0.2 0.2

0.02 0.02

180 0 180 0

0.015 0.015

0.01 0.01

210 330 210 330

0.005 0.005

240 300 240 300

270 270

0 0

(g) ψ̇ = −25 deg/sec (h) ψ̇ = 25 deg/sec

Figure 6.55: Elemental drag distribution for different turn rates at V = 60 kts.

293

0.04 0.04

90 90

1 1

120 60 120 60

0.8 0.035 0.8 0.035

150 30 150 30

0.4 0.4

0.025 0.025

0.2 0.2

0.02 0.02

180 0 180 0

0.015 0.015

0.01 0.01

210 330 210 330

0.005 0.005

0 0

240 300 240 300

270 270

−0.005 −0.005

(a) ψ̇ = 0 deg/sec (b) ψ̇ = 0 deg/sec

0.04 0.04

90 90

1 1

120 60 120 60

0.8 0.035 0.8 0.035

150 30 150 30

0.4 0.4

0.025 0.025

0.2 0.2

0.02 0.02

180 0 180 0

0.015 0.015

0.01 0.01

210 330 210 330

0.005 0.005

0 0

240 300 240 300

270 270

−0.005 −0.005

(c) ψ̇ = −5 deg/sec (d) ψ̇ = 5 deg/sec

0.04 0.04

90 90

1 1

120 60 120 60

0.8 0.035 0.8 0.035

150 30 150 30

0.4 0.4

0.025 0.025

0.2 0.2

0.02 0.02

180 0 180 0

0.015 0.015

0.01 0.01

210 330 210 330

0.005 0.005

0 0

240 300 240 300

270 270

−0.005 −0.005

(e) ψ̇ = −15 deg/sec (f) ψ̇ = 15 deg/sec

0.04 0.04

90 90

1 1

120 60 120 60

0.8 0.035 0.8 0.035

150 30 150 30

0.4 0.4

0.025 0.025

0.2 0.2

0.02 0.02

180 0 180 0

0.015 0.015

0.01 0.01

210 330 210 330

0.005 0.005

0 0

240 300 240 300

270 270

−0.005 −0.005

(g) ψ̇ = −25 deg/sec (h) ψ̇ = 25 deg/sec

Figure 6.56: Induced torque distribution for different turn rates at V = 60 kts.

294

0.03 0.03

90 90

1 1

120 60 120 60

0.8 0.8

0.025 0.025

0.6 0.6

150 30 150 30

0.4 0.4

0.02 0.02

0.2 0.2

0.01 0.01

210 330 210 330

0.005 0.005

270 270

0 0

(a) ψ̇ = 0 deg/sec (b) ψ̇ = 0 deg/sec

0.03 0.03

90 90

1 1

120 60 120 60

0.8 0.8

0.025 0.025

0.6 0.6

150 30 150 30

0.4 0.4

0.02 0.02

0.2 0.2

0.01 0.01

210 330 210 330

0.005 0.005

270 270

0 0

(c) ψ̇ = −5 deg/sec (d) ψ̇ = 5 deg/sec

0.03 0.03

90 90

1 1

120 60 120 60

0.8 0.8

0.025 0.025

0.6 0.6

150 30 150 30

0.4 0.4

0.02 0.02

0.2 0.2

0.01 0.01

210 330 210 330

0.005 0.005

270 270

0 0

(e) ψ̇ = −15 deg/sec (f) ψ̇ = 15 deg/sec

0.03 0.03

90 90

1 1

120 60 120 60

0.8 0.8

0.025 0.025

0.6 0.6

150 30 150 30

0.4 0.4

0.02 0.02

0.2 0.2

0.01 0.01

210 330 210 330

0.005 0.005

270 270

0 0

(g) ψ̇ = −25 deg/sec (h) ψ̇ = 25 deg/sec

Figure 6.57: Profile torque distribution for different turn rates at V = 60 kts.

295

0.01 0.01

90 90

1 1

120 60 120 60

0.8 0.005 0.8 0.005

0.6 0.6

150 30 0 150 30 0

0.4 0.4

−0.015 −0.015

−0.025 −0.025

240 300 240 300

270 270

−0.03 −0.03

(a) ψ̇ = 0 deg/sec (b) ψ̇ = 0 deg/sec

0.01 0.01

90 90

1 1

120 60 120 60

0.8 0.005 0.8 0.005

0.6 0.6

150 30 0 150 30 0

0.4 0.4

−0.015 −0.015

−0.025 −0.025

240 300 240 300

270 270

−0.03 −0.03

(c) ψ̇ = −5 deg/sec (d) ψ̇ = 5 deg/sec

0.01 0.01

90 90

1 1

120 60 120 60

0.8 0.005 0.8 0.005

0.6 0.6

150 30 0 150 30 0

0.4 0.4

−0.015 −0.015

−0.025 −0.025

240 300 240 300

270 270

−0.03 −0.03

(e) ψ̇ = −15 deg/sec (f) ψ̇ = 15 deg/sec

0.01 0.01

90 90

1 1

120 60 120 60

0.8 0.005 0.8 0.005

0.6 0.6

150 30 0 150 30 0

0.4 0.4

−0.015 −0.015

−0.025 −0.025

240 300 240 300

270 270

−0.03 −0.03

(g) ψ̇ = −25 deg/sec (h) ψ̇ = 25 deg/sec

Figure 6.58: Moment coefficient distribution for different turn rates at V = 60 kts.

296

−3 −3

x 10 x 10

5 5

90 90

1 1

120 60 120 60

0.8 0.8

0.6 0 0.6 0

150 30 150 30

0.4 0.4

0.2 −5 0.2 −5

180 0 180 0

−10 −10

−15 −15

270 270

−20 −20

(a) ψ̇ = 0 deg/sec −3

(b) ψ̇ = 0 deg/sec −3

x 10 x 10

5 5

90 90

1 1

120 60 120 60

0.8 0.8

0.6 0 0.6 0

150 30 150 30

0.4 0.4

0.2 −5 0.2 −5

180 0 180 0

−10 −10

−15 −15

270 270

−20 −20

(c) ψ̇ = −5 deg/sec −3

(d) ψ̇ = 5 deg/sec −3

x 10 x 10

5 5

90 90

1 1

120 60 120 60

0.8 0.8

0.6 0 0.6 0

150 30 150 30

0.4 0.4

0.2 −5 0.2 −5

180 0 180 0

−10 −10

−15 −15

270 270

−20 −20

(e) ψ̇ = −15 deg/sec −3

(f) ψ̇ = 15 deg/sec −3

x 10 x 10

5 5

90 90

1 1

120 60 120 60

0.8 0.8

0.6 0 0.6 0

150 30 150 30

0.4 0.4

0.2 −5 0.2 −5

180 0 180 0

−10 −10

−15 −15

270 270

−20 −20

(g) ψ̇ = −25 deg/sec (h) ψ̇ = 25 deg/sec

Figure 6.59: Elemental moment distribution for different turn rates at V = 60 kts.

297

9

25 deg/sec

8 15 deg/sec

5 deg/sec

Equivalent flap angle, β (deg)

7 level

–5 deg/sec

6 –15 deg/sec

–25 deg/sec

5

1

0 90 180 270 360

Azimuth angle, ψ (deg)

298

2.0

1st harmonic

2nd harmonic

3rd harmonic

1.5

1.0

0.5

0.0

-25 -15 -5 0 5 15 25

(a) Magnitude.

180

1st harmonic

2nd harmonic

3rd harmonic

Phase of flap harmonics (deg)

90

-90

-180

-25 -15 -5 0 5 15 25

(b) Phase.

Figure 6.61: Magnitude and phase of the first three flap harmonics at various turn

rates; V = 60 kts.

299

0

25 deg/sec –5 deg/sec

15 deg/sec –15 deg/sec

-2 5 deg/sec –25 deg/sec

Equivalent lag angle, ζ (deg)

level

-4

-6

-8

-10

Azimuth angle, ψ (deg)

300

0.6

1st harmonic

2nd harmonic

0.5 3rd harmonic

0.4

0.3

0.2

0.1

0.0

-25 -15 -5 0 5 15 25

(a) Magnitude.

180

1st harmonic

2nd harmonic

3rd harmonic

Phase of lag harmonics (deg)

90

-90

-180

-25 -15 -5 0 5 15 25

(b) Phase.

Figure 6.63: Magnitude and phase of the first three lag harmonics at various turn

rates; V = 60 kts.

301

2.0

Elastic torsion at the blade tip, φ (deg)

1.5

1.0

0.5

0.0

-0.5

15 deg/sec –15 deg/sec

-1.5 5 deg/sec –25 deg/sec

level

-2.0

0 90 180 270 360

Azimuth angle, ψ (deg)

Figure 6.64: Blade tip elastic torsion versus azimuth angle at various turn rates;

V = 60 kts.

302

1.0

1st harmonic

2nd harmonic

3rd harmonic

0.8

0.6

0.4

0.2

0.0

-25 -15 -5 0 5 15 25

(a) Magnitude.

180

Phase of torsion harmonics (deg)

90

-90

1st harmonic

2nd harmonic

3rd harmonic

-180

-25 -15 -5 0 5 15 25

(b) Phase.

Figure 6.65: Magnitude and phase of the first three torsion harmonics at various

303

Flight path angle, γ (deg)

-20 -10 0 10 20 30 40

Main Rotor Power Required (HP) 3000

Simulation

2000

1500

1000

500

-500

-3000 -2000 -1000 0 1000 2000 3000 4000

Rate of climb/descent, Vc (ft/min)

-20 -10 0 10 20 30 40

100

Main Rotor Collective Stick, , δcol (%)

Flight test

80 Simulation

60

40

20

0

-3000 -2000 -1000 0 1000 2000 3000 4000

Rate of climb/descent, Vc (ft/min)

Figure 6.66: Main rotor power required, QM R , and collective stick position, δcol , as

304

Flight path angle, γ (deg)

-20 -10 0 10 20 30 40

Longitudinal Stick Position, δlon (%) 100

Flight test

80 Simulation

60

Aft

Forward

40

20

0

-3000 -2000 -1000 0 1000 2000 3000 4000

Rate of climb/descent, Vc (ft/min)

-20 -10 0 10 20 30 40

100

Flight test

Lateral Stick Position, δlat (%)

80 Simulation

Right

60

40

Left

20

0

-3000 -2000 -1000 0 1000 2000 3000 4000

Rate of climb/descent, Vc (ft/min)

Figure 6.67: Longitudinal stick position, δlon , and lateral stick position, δlat , as a

305

Flight path angle, γ (deg)

-20 -10 0 10 20 30 40

5

4 Flight test

Simulation

Pitch attitude, θ (deg)

3

2

1

0

-1

-2

-3

-3000 -2000 -1000 0 1000 2000 3000 4000

Rate of climb/descent, Vc (ft/min)

-20 -10 0 10 20 30 40

100

Flight test

Simulation

Pedal Position, δped (%)

80

Right

60

40

Left

20

0

-3000 -2000 -1000 0 1000 2000 3000 4000

Rate of climb/descent, Vc (ft/min)

Figure 6.68: Helicopter pitch attitude, θ, and pedal setting, δped ,as a function of

306

1.5 1.5

(a) γ = 0◦ (b) γ = 9◦

1.0 1.0

0.5 0.5

z/R

z/R

0.0 0.0

−0.5 −0.5

−1.0 −1.0

−1.5 −1.5

−1 0 1 2 3 −1 0 1 2 3

x/R x/R

1.5 1.5

(c) γ = 3◦ (d) γ = 12◦

1.0 1.0

0.5 0.5

z/R

z/R

0.0 0.0

−0.5 −0.5

−1.0 −1.0

−1.5 −1.5

−1 0 1 2 3 −1 0 1 2 3

x/R x/R

1.5 1.5

(e) γ = 6◦ (f) γ = 18◦

1.0 1.0

0.5 0.5

z/R

z/R

0.0 0.0

−0.5 −0.5

−1.0 −1.0

−1.5 −1.5

−1 0 1 2 3 −1 0 1 2 3

x/R x/R

Figure 6.69: Side view of the wake geometry for different climb angles γ; V = 60

kts.

307

1.5 1.5

(a) γ = 0◦ (b) γ = 9◦

1.0 1.0

0.5 0.5

z/R

z/R

0.0 0.0

−0.5 −0.5

−1.0 −1.0

−1.5 −1.5

−1.5 −1.0 −0.5 0.0 0.5 1.0 1.5 −1.5 −1.0 −0.5 0.0 0.5 1.0 1.5

y/R y/R

1.5 1.5

(c) γ = 3◦ (d) γ = 12◦

1.0 1.0

0.5 0.5

z/R

z/R

0.0 0.0

−0.5 −0.5

−1.0 −1.0

−1.5 −1.5

−1.5 −1.0 −0.5 0.0 0.5 1.0 1.5 −1.5 −1.0 −0.5 0.0 0.5 1.0 1.5

y/R y/R

1.5 1.5

(e) γ = 6◦ (f) γ = 18◦

1.0 1.0

0.5 0.5

z/R

z/R

0.0 0.0

−0.5 −0.5

−1.0 −1.0

−1.5 −1.5

−1.5 −1.0 −0.5 0.0 0.5 1.0 1.5 −1.5 −1.0 −0.5 0.0 0.5 1.0 1.5

y/R y/R

Figure 6.70: Rear view of the wake geometry for different climb angles γ; V = 60

kts.

308

1.5 1.5

1.0 1.0

0.5 0.5

y/R

y/R

0.0 0.0

−0.5 −0.5

−1.0 −1.0

(a) γ = 0◦ (b) γ = 9◦

−1.5 −1.5

−1 0 1 2 3 −1 0 1 2 3

x/R x/R

1.5 1.5

1.0 1.0

0.5 0.5

y/R

y/R

0.0 0.0

−0.5 −0.5

−1.0 −1.0

◦

(c) γ = 3 (d) γ = 12◦

−1.5 −1.5

−1 0 1 2 3 −1 0 1 2 3

x/R x/R

1.5 1.5

1.0 1.0

0.5 0.5

y/R

y/R

0.0 0.0

−0.5 −0.5

−1.0 −1.0

◦

(e) γ = 6 (f) γ = 18◦

−1.5 −1.5

−1 0 1 2 3 −1 0 1 2 3

x/R x/R

Figure 6.71: Top view of the wake geometry for different climb angles γ; V = 60

kts.

309

1.5 1.5

1.0 1.0

0.5 0.5

z/R

z/R

0.0 0.0

−0.5 −0.5

−1.0 −1.0

−1.5 −1.5

−1 0 1 2 3 −1 0 1 2 3

x/R x/R

1.5 1.5

1.0 1.0

0.5 0.5

z/R

z/R

0.0 0.0

−0.5 −0.5

−1.0 −1.0

◦

(c) γ = −3 (d) γ = −12◦

−1.5 −1.5

−1 0 1 2 3 −1 0 1 2 3

x/R x/R

1.5 1.5

1.0 1.0

0.5 0.5

z/R

z/R

0.0 0.0

−0.5 −0.5

−1.0 −1.0

◦

(e) γ = −6 (f) γ = −18◦

−1.5 −1.5

−1 0 1 2 3 −1 0 1 2 3

x/R x/R

Figure 6.72: Side view of the wake geometry for different descent angles γ; V = 60

kts.

310

1.5 1.5

1.0 1.0

0.5 0.5

z/R

z/R

0.0 0.0

−0.5 −0.5

−1.0 −1.0

−1.5 −1.5

−1.5 −1.0 −0.5 0.0 0.5 1.0 1.5 −1.5 −1.0 −0.5 0.0 0.5 1.0 1.5

y/R y/R

1.5 1.5

1.0 1.0

0.5 0.5

z/R

z/R

0.0 0.0

−0.5 −0.5

−1.0 −1.0

◦

(c) γ = −3 (d) γ = −12◦

−1.5 −1.5

−1.5 −1.0 −0.5 0.0 0.5 1.0 1.5 −1.5 −1.0 −0.5 0.0 0.5 1.0 1.5

y/R y/R

1.5 1.5

1.0 1.0

0.5 0.5

z/R

z/R

0.0 0.0

−0.5 −0.5

−1.0 −1.0

−1.5 −1.5

−1.5 −1.0 −0.5 0.0 0.5 1.0 1.5 −1.5 −1.0 −0.5 0.0 0.5 1.0 1.5

y/R y/R

Figure 6.73: Rear view of the wake geometry for different descent angles γ; V = 60

kts.

311

1.5 1.5

1.0 1.0

0.5 0.5

y/R

y/R

0.0 0.0

−0.5 −0.5

−1.0 −1.0

−1.5 −1.5

−1 0 1 2 3 −1 0 1 2 3

x/R x/R

1.5 1.5

1.0 1.0

0.5 0.5

y/R

y/R

0.0 0.0

−0.5 −0.5

−1.0 −1.0

◦

(c) γ = −3 (d) γ = −12◦

−1.5 −1.5

−1 0 1 2 3 −1 0 1 2 3

x/R x/R

1.5 1.5

1.0 1.0

0.5 0.5

y/R

y/R

0.0 0.0

−0.5 −0.5

−1.0 −1.0

◦

(e) γ = −6 (f) γ = −18◦

−1.5 −1.5

−1 0 1 2 3 −1 0 1 2 3

x/R x/R

Figure 6.74: Top view of the wake geometry for different descent angles γ; V = 60

kts.

312

0.15 0.15

90 90

1 1

120 60 120 60

0.8 0.8

0.6 0.6

150 30 0.1 150 30 0.1

0.4 0.4

0.2 0.2

270 270

−0.05 −0.05

(a) γ = 0◦ (b) γ = 9◦

0.15 0.15

90 90

1 1

120 60 120 60

0.8 0.8

0.6 0.6

150 30 0.1 150 30 0.1

0.4 0.4

0.2 0.2

270 270

−0.05 −0.05

(c) γ = 3◦ (d) γ = 12◦

0.15 0.15

90 90

1 1

120 60 120 60

0.8 0.8

0.6 0.6

150 30 0.1 150 30 0.1

0.4 0.4

0.2 0.2

270 270

−0.05 −0.05

(e) γ = 6◦ (f) γ = 18◦

313

0.15 0.15

90 90

1 1

120 60 120 60

0.8 0.8

0.6 0.6

150 30 0.1 150 30 0.1

0.4 0.4

0.2 0.2

270 270

−0.05 −0.05

(a) γ = 0◦ (b) γ = −9◦

0.15 0.15

90 90

1 1

120 60 120 60

0.8 0.8

0.6 0.6

150 30 0.1 150 30 0.1

0.4 0.4

0.2 0.2

270 270

−0.05 −0.05

(c) γ = −3◦ (d) γ = −12◦

0.15 0.15

90 90

1 1

120 60 120 60

0.8 0.8

0.6 0.6

150 30 0.1 150 30 0.1

0.4 0.4

0.2 0.2

270 270

−0.05 −0.05

(e) γ = −6◦ (f) γ = −18◦

314

15 15

90 90

1 1

120 60 120 60

0.8 0.8

0.6 0.6

150 30 10 150 30 10

0.4 0.4

0.2 0.2

180 0 5 180 0 5

270 270

−5 −5

(a) γ = 0◦ (b) γ = 9◦

15 15

90 90

1 1

120 60 120 60

0.8 0.8

0.6 0.6

150 30 10 150 30 10

0.4 0.4

0.2 0.2

180 0 5 180 0 5

270 270

−5 −5

(c) γ = 3◦ (d) γ = 12◦

15 15

90 90

1 1

120 60 120 60

0.8 0.8

0.6 0.6

150 30 10 150 30 10

0.4 0.4

0.2 0.2

180 0 5 180 0 5

270 270

−5 −5

(e) γ = 6◦ (f) γ = 18◦

Figure 6.77: Angle of attack distribution for different climb angles γ; V = 60 kts.

315

15 15

90 90

1 1

120 60 120 60

0.8 0.8

0.6 0.6

150 30 10 150 30 10

0.4 0.4

0.2 0.2

180 0 5 180 0 5

270 270

−5 −5

(a) γ = 0◦ (b) γ = −9◦

15 15

90 90

1 1

120 60 120 60

0.8 0.8

0.6 0.6

150 30 10 150 30 10

0.4 0.4

0.2 0.2

180 0 5 180 0 5

270 270

−5 −5

(c) γ = −3◦ (d) γ = −12◦

15 15

90 90

1 1

120 60 120 60

0.8 0.8

0.6 0.6

150 30 10 150 30 10

0.4 0.4

0.2 0.2

180 0 5 180 0 5

270 270

−5 −5

(e) γ = −6◦ (f) γ = −18◦

Figure 6.78: Angle of attack distribution for different descent angles γ; V = 60 kts.

316

90 90

1 1

120 60 1 120 60 1

0.8 0.8

0.8 0.8

0.6 0.6

150 30 0.6 150 30 0.6

0.4 0.4

0.4 0.4

0.2 0.2

0.2 0.2

180 0 180 0

0 0

−0.2 −0.2

−0.6 −0.6

270 270

−1 −1

(a) γ = 0◦ (b) γ = 9◦

90 90

1 1

120 60 1 120 60 1

0.8 0.8

0.8 0.8

0.6 0.6

150 30 0.6 150 30 0.6

0.4 0.4

0.4 0.4

0.2 0.2

0.2 0.2

180 0 180 0

0 0

−0.2 −0.2

−0.6 −0.6

270 270

−1 −1

(c) γ = 3◦ (d) γ = 12◦

90 90

1 1

120 60 1 120 60 1

0.8 0.8

0.8 0.8

0.6 0.6

150 30 0.6 150 30 0.6

0.4 0.4

0.4 0.4

0.2 0.2

0.2 0.2

180 0 180 0

0 0

−0.2 −0.2

−0.6 −0.6

270 270

−1 −1

(e) γ = 6◦ (f) γ = 18◦

Figure 6.79: Lift coefficient distribution for different climb angles γ; V = 60 kts.

317

90 90

1 1

120 60 1 120 60 1

0.8 0.8

0.8 0.8

0.6 0.6

150 30 0.6 150 30 0.6

0.4 0.4

0.4 0.4

0.2 0.2

0.2 0.2

180 0 180 0

0 0

−0.2 −0.2

−0.6 −0.6

270 270

−1 −1

(a) γ = 0◦ (b) γ = −9◦

90 90

1 1

120 60 1 120 60 1

0.8 0.8

0.8 0.8

0.6 0.6

150 30 0.6 150 30 0.6

0.4 0.4

0.4 0.4

0.2 0.2

0.2 0.2

180 0 180 0

0 0

−0.2 −0.2

−0.6 −0.6

270 270

−1 −1

(c) γ = −3◦ (d) γ = −12◦

90 90

1 1

120 60 1 120 60 1

0.8 0.8

0.8 0.8

0.6 0.6

150 30 0.6 150 30 0.6

0.4 0.4

0.4 0.4

0.2 0.2

0.2 0.2

180 0 180 0

0 0

−0.2 −0.2

−0.6 −0.6

270 270

−1 −1

(e) γ = −6◦ (f) γ = −18◦

Figure 6.80: Lift coefficient distribution for different descent angles γ; V = 60 kts.

318

90 0.2 90 0.2

1 1

120 60 120 60

0.8 0.8

0.15 0.15

0.6 0.6

150 30 150 30

0.4 0.1 0.4 0.1

0.2 0.2

0.05 0.05

180 0 180 0

0 0

−0.1 −0.1

240 300 240 300

270 270

(a) γ = 0◦ (b) γ = 9◦

90 0.2 90 0.2

1 1

120 60 120 60

0.8 0.8

0.15 0.15

0.6 0.6

150 30 150 30

0.4 0.1 0.4 0.1

0.2 0.2

0.05 0.05

180 0 180 0

0 0

−0.1 −0.1

240 300 240 300

270 270

(c) γ = 3◦ (d) γ = 12◦

90 0.2 90 0.2

1 1

120 60 120 60

0.8 0.8

0.15 0.15

0.6 0.6

150 30 150 30

0.4 0.1 0.4 0.1

0.2 0.2

0.05 0.05

180 0 180 0

0 0

−0.1 −0.1

240 300 240 300

270 270

(e) γ = 6◦ (f) γ = 18◦

Figure 6.81: Elemental lift distribution for different climb angles γ; V = 60 kts.

319

90 0.2 90 0.2

1 1

120 60 120 60

0.8 0.8

0.15 0.15

0.6 0.6

150 30 150 30

0.4 0.1 0.4 0.1

0.2 0.2

0.05 0.05

180 0 180 0

0 0

−0.1 −0.1

240 300 240 300

270 270

(a) γ = 0◦ (b) γ = −9◦

90 0.2 90 0.2

1 1

120 60 120 60

0.8 0.8

0.15 0.15

0.6 0.6

150 30 150 30

0.4 0.1 0.4 0.1

0.2 0.2

0.05 0.05

180 0 180 0

0 0

−0.1 −0.1

240 300 240 300

270 270

(c) γ = −3◦ (d) γ = −12◦

90 0.2 90 0.2

1 1

120 60 120 60

0.8 0.8

0.15 0.15

0.6 0.6

150 30 150 30

0.4 0.1 0.4 0.1

0.2 0.2

0.05 0.05

180 0 180 0

0 0

−0.1 −0.1

240 300 240 300

270 270

(e) γ = −6◦ (f) γ = −18◦

Figure 6.82: Elemental lift distribution for different descent angles γ; V = 60 kts.

320

90 90

1 1

120 60 0.15 120 60 0.15

0.8 0.8

150 30 150 30

0.4 0.4

0.05 0.05

0.2 0.2

180 0 0 180 0 0

−0.05 −0.05

−0.1 −0.1

270 270

(a) γ = 0◦ (b) γ = 9◦

90 90

1 1

120 60 0.15 120 60 0.15

0.8 0.8

150 30 150 30

0.4 0.4

0.05 0.05

0.2 0.2

180 0 0 180 0 0

−0.05 −0.05

−0.1 −0.1

270 270

(c) γ = 3◦ (d) γ = 12◦

90 90

1 1

120 60 0.15 120 60 0.15

0.8 0.8

150 30 150 30

0.4 0.4

0.05 0.05

0.2 0.2

180 0 0 180 0 0

−0.05 −0.05

−0.1 −0.1

270 270

(e) γ = 6◦ (f) γ = 18◦

Figure 6.83: Local flap moment for different climb angles γ; V = 60 kts.

321

90 90

1 1

120 60 0.15 120 60 0.15

0.8 0.8

150 30 150 30

0.4 0.4

0.05 0.05

0.2 0.2

180 0 0 180 0 0

−0.05 −0.05

−0.1 −0.1

270 270

(a) γ = 0◦ (b) γ = −9◦

90 90

1 1

120 60 0.15 120 60 0.15

0.8 0.8

150 30 150 30

0.4 0.4

0.05 0.05

0.2 0.2

180 0 0 180 0 0

−0.05 −0.05

−0.1 −0.1

270 270

(c) γ = −3◦ (d) γ = −12◦

90 90

1 1

120 60 0.15 120 60 0.15

0.8 0.8

150 30 150 30

0.4 0.4

0.05 0.05

0.2 0.2

180 0 0 180 0 0

−0.05 −0.05

−0.1 −0.1

270 270

(e) γ = −6◦ (f) γ = −18◦

Figure 6.84: Local flap moment for different descent angles γ; V = 60 kts.

322

0.03 0.03

90 90

1 1

120 60 0.028 120 60 0.028

0.8 0.8

150 30 150 30

0.4 0.024 0.4 0.024

0.018 0.018

0.016 0.016

210 330 210 330

0.014 0.014

270 270

0.01 0.01

(a) γ = 0◦ (b) γ = 9◦

0.03 0.03

90 90

1 1

120 60 0.028 120 60 0.028

0.8 0.8

150 30 150 30

0.4 0.024 0.4 0.024

0.018 0.018

0.016 0.016

210 330 210 330

0.014 0.014

270 270

0.01 0.01

(c) γ = 3◦ (d) γ = 12◦

0.03 0.03

90 90

1 1

120 60 0.028 120 60 0.028

0.8 0.8

150 30 150 30

0.4 0.024 0.4 0.024

0.018 0.018

0.016 0.016

210 330 210 330

0.014 0.014

270 270

0.01 0.01

(e) γ = 6◦ (f) γ = 18◦

Figure 6.85: Drag coefficient distribution for different climb angles γ; V = 60 kts.

323

0.03 0.03

90 90

1 1

120 60 0.028 120 60 0.028

0.8 0.8

150 30 150 30

0.4 0.024 0.4 0.024

0.018 0.018

0.016 0.016

## Гораздо больше, чем просто документы.

Откройте для себя все, что может предложить Scribd, включая книги и аудиокниги от крупных издательств.

Отменить можно в любой момент.